#LyX 2.3 created this file. For more info see http://www.lyx.org/ \lyxformat 544 \begin_document \begin_header \save_transient_properties true \origin unavailable \textclass extbook \begin_preamble % List all used files in log output \listfiles % Add a DRAFT watermark \usepackage{draftwatermark} \usepackage{accsupp} \SetWatermarkLightness{0.97} \SetWatermarkScale{1} % Make watermark not copyable (in Adobe Reader) \SetWatermarkText{\BeginAccSupp{method=escape,ActualText={}}DRAFT\EndAccSupp{}} % Set up required header format \usepackage{fancyhdr} \pagestyle{fancy} \renewcommand{\headrulewidth}{0pt} \rhead{} \lhead{} \chead{} \rfoot{} \lfoot{} % Make page number not copyable (in Adobe Reader) \cfoot{\BeginAccSupp{method=escape,ActualText={}}\thepage\EndAccSupp{}} % Page number bottom center % Allow FloatBarrier command \usepackage{placeins} % Allow landscape pages \usepackage{pdflscape} % Allow doing things after the end of the current page % (to avoid landscape figures breaking up text) \usepackage{afterpage} % This one breaks subfigs so it's disabled % https://tex.stackexchange.com/questions/65680/automatically-bold-first-sentence-of-a-floats-caption \end_preamble \use_default_options true \begin_modules todonotes logicalmkup \end_modules \maintain_unincluded_children false \language english \language_package default \inputencoding utf8 \fontencoding default \font_roman "default" "default" \font_sans "default" "default" \font_typewriter "default" "default" \font_math "auto" "auto" \font_default_family default \use_non_tex_fonts false \font_sc false \font_osf false \font_sf_scale 100 100 \font_tt_scale 100 100 \use_microtype false \use_dash_ligatures true \graphics default \default_output_format pdf4 \output_sync 0 \bibtex_command biber \index_command default \paperfontsize 12 \spacing double \use_hyperref true \pdf_bookmarks true \pdf_bookmarksnumbered false \pdf_bookmarksopen false \pdf_bookmarksopenlevel 1 \pdf_breaklinks false \pdf_pdfborder false \pdf_colorlinks false \pdf_backref false \pdf_pdfusetitle true \papersize letterpaper \use_geometry true \use_package amsmath 1 \use_package amssymb 1 \use_package cancel 1 \use_package esint 1 \use_package mathdots 1 \use_package mathtools 1 \use_package mhchem 1 \use_package stackrel 1 \use_package stmaryrd 1 \use_package undertilde 1 \cite_engine biblatex \cite_engine_type authoryear \biblio_style plain \biblatex_bibstyle authoryear \biblatex_citestyle numeric \use_bibtopic false \use_indices false \paperorientation portrait \suppress_date false \justification true \use_refstyle 1 \use_minted 0 \index Index \shortcut idx \color #008000 \end_index \leftmargin 1.5in \topmargin 1in \rightmargin 1in \bottommargin 1in \secnumdepth 3 \tocdepth 3 \paragraph_separation indent \paragraph_indentation default \is_math_indent 0 \math_numbering_side default \quotes_style english \dynamic_quotes 0 \papercolumns 1 \papersides 1 \paperpagestyle default \tracking_changes false \output_changes false \html_math_output 0 \html_css_as_file 0 \html_be_strict false \end_header \begin_body \begin_layout Title Bioinformatic analysis of complex, high-throughput genomic and epigenomic data in the context of immunology and transplant rejection \end_layout \begin_layout Author A thesis presented \begin_inset Newline newline \end_inset by \begin_inset Newline newline \end_inset Ryan C. Thompson \begin_inset Newline newline \end_inset to \begin_inset Newline newline \end_inset The Scripps Research Institute Graduate Program \begin_inset Newline newline \end_inset in partial fulfillment of the requirements for the degree of \begin_inset Newline newline \end_inset Doctor of Philosophy in the subject of Biology \begin_inset Newline newline \end_inset for \begin_inset Newline newline \end_inset The Scripps Research Institute \begin_inset Newline newline \end_inset La Jolla, California \end_layout \begin_layout Date October 2019 \end_layout \begin_layout Standard [Copyright notice] \end_layout \begin_layout Standard [Thesis acceptance form] \end_layout \begin_layout Standard [Dedication] \end_layout \begin_layout Standard [Acknowledgements] \end_layout \begin_layout Standard \begin_inset CommandInset toc LatexCommand tableofcontents \end_inset \end_layout \begin_layout Standard \begin_inset FloatList table \end_inset \end_layout \begin_layout Standard \begin_inset FloatList figure \end_inset \end_layout \begin_layout Standard [List of Abbreviations] \end_layout \begin_layout List of TODOs \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Check all figures to make sure they fit on the page with their legends. \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Search and replace: naive -> naïve \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Look into auto-generated nomenclature list: \begin_inset CommandInset href LatexCommand href target "https://wiki.lyx.org/Tips/Nomenclature" \end_inset . Otherwise, do a manual pass for all abbreviations at the end. Do nomenclature/abbreviations independently for each chapter. \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Make all descriptions consistent in terms of \begin_inset Quotes eld \end_inset we did X \begin_inset Quotes erd \end_inset vs \begin_inset Quotes eld \end_inset I did X \begin_inset Quotes erd \end_inset vs \begin_inset Quotes eld \end_inset X was done \begin_inset Quotes erd \end_inset . \end_layout \end_inset \end_layout \begin_layout Chapter* Abstract \end_layout \begin_layout Standard \begin_inset Note Note status open \begin_layout Plain Layout It is included as an integral part of the thesis and should immediately precede the introduction. \end_layout \begin_layout Plain Layout Preparing your Abstract. Your abstract (a succinct description of your work) is limited to 350 words. UMI will shorten it if they must; please do not exceed the limit. \end_layout \begin_layout Itemize Include pertinent place names, names of persons (in full), and other proper nouns. These are useful in automated retrieval. \end_layout \begin_layout Itemize Display symbols, as well as foreign words and phrases, clearly and accurately. Include transliterations for characters other than Roman and Greek letters and Arabic numerals. Include accents and diacritical marks. \end_layout \begin_layout Itemize Do not include graphs, charts, tables, or illustrations in your abstract. \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Obviously the abstract gets written last. \end_layout \end_inset \end_layout \begin_layout Chapter* Notes to draft readers \end_layout \begin_layout Standard Thank you so much for agreeing to read my thesis and give me feedback on it. What you are currently reading is a rough draft, in need of many revisions. You can always find the latest version at \begin_inset CommandInset href LatexCommand href target "https://mneme.dedyn.io/~ryan/Thesis/thesis.pdf" literal "false" \end_inset . the PDF at this link is updated periodically with my latest revisions, but you can just download the current version and give me feedback on that. Don't worry about keeping up with the updates. \end_layout \begin_layout Standard As for what feedback I'm looking for, first of all, don't waste your time marking spelling mistakes and such. I haven't run a spell checker on it yet, so let me worry about that. Also, I'm aware that many abbreviations are not properly introduced the first time they are used, so don't worry about that either. However, if you see any glaring formatting issues, such as a figure being too large and getting cut off at the edge of the page, please note them. In addition, if any of the text in the figures is too small, please note that as well. \end_layout \begin_layout Standard Beyond that, what I'm mainly interested in is feedback on the content. For example: does the introduction flow logically, and does it provide enough background to understand the other chapters? Does each chapter make it clear what work and analyses I have done? Do the figures clearly communicate the results I'm trying to show? Do you feel that the claims in the results and discussion sections are well-supported? There's no need to suggest improvements; just note areas that you feel need improvement. Additionally, while I am well aware that Chapter 1 (the introduction) contains many un-cited claims, all the other chapters (2,3, and 4) \emph on should \emph default be fully cited. So if you notice any un-cited claims in those chapters, please flag them for my attention. Similarly, if you discover any factual errors, please note them as well. \end_layout \begin_layout Standard You can provide your feedback in whatever way is most convenient to you. You could mark up this PDF with highlights and notes, then send it back to me. Or you could collect your comments in a separate text file and send that to me, or whatever else you like. However, if you send me your feedback in a separate document, please note a section/figure/table number for each comment, and \emph on also \emph default send me the exact PDF that you read so I can reference it while reading your comments, since as mentioned above, the current version I'm working on will have changed by that point (which might include shuffling sections and figures around, changing their numbers). One last thing: you'll see a bunch of text in orange boxes throughout the PDF. These are notes to myself about things that need to be fixed later, so if you see a problem noted in an orange box, that means I'm already aware of it, and there's no need to comment on it. \end_layout \begin_layout Standard My thesis is due Thursday, October 10th, so in order to be useful to me, I'll need your feedback at least a few days before that, ideally by Monday, October 7th. If you have limited time and are unable to get through the whole thesis, please focus your efforts on Chapters 1 and 2, since those are the roughest and most in need of revision. Chapter 3 is fairly short and straightforward, and Chapter 4 is an adaptation of a paper that's already been through a few rounds of revision, so they should be a lot tighter. If you can't spare any time between now and then, or if something unexpected comes up, I understand. Just let me know. \end_layout \begin_layout Standard Thanks again for your help, and happy reading! \end_layout \begin_layout Chapter Introduction \end_layout \begin_layout Section Background & Significance \end_layout \begin_layout Subsection Biological motivation \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Rethink the subsection organization after the intro is written. \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Citations are needed all over the place. A lot of this is knowledge I've just absorbed from years of conversation in the Salomon lab, without ever having seen a citation for it. \end_layout \end_inset \end_layout \begin_layout Subsubsection Rejection is the major long-term threat to organ and tissue allografts \end_layout \begin_layout Standard Organ and tissue transplants are a life-saving treatment for people who have lost the function of an important organ [CITE?]. In some cases, it is possible to transplant a patient's own tissue from one area of their body to another, referred to as an autograft. This is common for tissues that are distributed throughout many areas of the body, such as skin and bone. However, in cases of organ failure, there is no functional self tissue remaining, and a transplant from another person – a donor – is required. This is referred to as an allograft. \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Possible citation for degree of generic variability: \begin_inset CommandInset href LatexCommand href target "https://www.ncbi.nlm.nih.gov/pubmed/22424236?dopt=Abstract" \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout How much mechanistic detail is needed here? My work doesn't really go into specific rejection mechanisms, so I think it's best to keep it basic. \end_layout \end_inset \end_layout \begin_layout Standard Because an allograft comes from a different person, it is genetically distinct from the rest of the recipient's body. Some genetic variants occur in protein coding regions, resulting in protein products that differ from the equivalent proteins in the graft recipient's own tissue. As a result, without intervention, the recipient's immune system will eventuall y identify the graft as foreign tissue and begin attacking it, eventually resulting in failure and death of the graft, a process referred to as transplan t rejection. Rejection is the most significant challenge to the long-term health and survival of an allograft [CITE?]. Like any adaptive immune response, graft rejection generally occurs via two broad mechanisms: cellular immunity, in which CD8+ T-cells recognizing graft-specific antigens induce apoptosis in the graft cells; and humoral immunity, in which B-cells produce antibodies that bind to graft proteins and direct an immune response against the graft [CITE?]. In either case, rejection shows most of the typical hallmarks of an adaptive immune response, in particular mediation by CD4+ T-cells and formation of immune memory. \end_layout \begin_layout Subsubsection Diagnosis and treatment of allograft rejection is a major challenge \end_layout \begin_layout Standard To prevent rejection, allograft recipients are treated with immune suppressive drugs [CITE?]. The goal is to achieve sufficient suppression of the immune system to prevent rejection of the graft without compromising the ability of the immune system to raise a normal response against infection. As such, a delicate balance must be struck: insufficient immune suppression may lead to rejection and ultimately loss of the graft; excessive suppression leaves the patient vulnerable to life-threatening opportunistic infections. Because every patient is different, immune suppression must be tailored for each patient. Furthermore, immune suppression must be tuned over time, as the immune system's activity is not static, nor is it held in a steady state [CITE?]. In order to properly adjust the dosage of immune suppression drugs, it is necessary to monitor the health of the transplant and increase the dosage if evidence of rejection is observed. \end_layout \begin_layout Standard However, diagnosis of rejection is a significant challenge. Early diagnosis is essential in order to step up immune suppression before the immune system damages the graft beyond recovery [CITE?]. The current gold standard test for graft rejection is a tissue biopsy, examined for visible signs of rejection by a trained histologist [CITE?]. When a patient shows symptoms of possible rejection, a \begin_inset Quotes eld \end_inset for cause \begin_inset Quotes erd \end_inset biopsy is performed to confirm the diagnosis, and immune suppression is adjusted as necessary. However, in many cases, the early stages of rejection are asymptomatic, known as \begin_inset Quotes eld \end_inset sub-clinical \begin_inset Quotes erd \end_inset rejection [CITE?]. In light of this, is is now common to perform \begin_inset Quotes eld \end_inset protocol biopsies \begin_inset Quotes erd \end_inset at specific times after transplantation of a graft, even if no symptoms of rejection are apparent, in addition to \begin_inset Quotes eld \end_inset for cause \begin_inset Quotes erd \end_inset biopsies \begin_inset CommandInset citation LatexCommand cite key "Wilkinson2006" literal "false" \end_inset . \end_layout \begin_layout Standard However, biopsies have a number of downsides that limit their effectiveness as a diagnostic tool. First, the need for manual inspection by a histologist means that diagnosis is subject to the biases of the particular histologist examining the biopsy [CITE?]. In marginal cases, two different histologists may give two different diagnoses to the same biopsy. Second, a biopsy can only evaluate if rejection is occurring in the section of the graft from which the tissue was extracted. If rejection is localized to one section of the graft and the tissue is extracted from a different section, a false negative diagnosis may result. Most importantly, extraction of tissue from a graft is invasive and is treated as an injury by the body, which results in inflammation that in turn promotes increased immune system activity [CITE?]. Hence, the invasiveness of biopsies severely limits the frequency with which they can safely be performed. Typically, protocol biopsies are not scheduled more than about once per month \begin_inset CommandInset citation LatexCommand cite key "Wilkinson2006" literal "false" \end_inset . A less invasive diagnostic test for rejection would bring manifold benefits. Such a test would enable more frequent testing and therefore earlier detection of rejection events. In addition, having a larger pool of historical data for a given patient would make it easier to evaluate when a given test is outside the normal parameters for that specific patient, rather than relying on normal ranges for the population as a whole. Lastly, the accumulated data from more frequent tests would be a boon to the transplant research community. Beyond simply providing more data overall, the better time granularity of the tests will enable studying the progression of a rejection event on the scale of days to weeks, rather than months. \end_layout \begin_layout Subsubsection Memory cells are resistant to immune suppression \end_layout \begin_layout Standard One of the defining features of the adaptive immune system is immune memory: the ability of the immune system to recognize a previously encountered foreign antigen and respond more quickly and more strongly to that antigen in subsequent encounters. When the immune system first encounters a new antigen, the lymphocytes that respond are known as naive cells – T-cells and B-cells that have never detected their target antigens before. Once activated by their specific antigen presented by an antigen-presenting cell in the proper co-stimulatory context, naive cells differentiate into effector cells that carry out their respective functions in targeting and destroying the source of the foreign antigen. The requirement for co-stimulation is an important feature of naive cells that limits \begin_inset Quotes eld \end_inset false positive \begin_inset Quotes erd \end_inset immune responses, because antigen-presenting cells usually only express the proper co-stimulation after detecting evidence of an infection, such as the presence of common bacterial cell components or inflamed tissue. Most effector cells die after the foreign antigen is cleared, since they are no longer needed, but some remain and differentiate into memory cells. Like naive cells, memory cells respond to detection of their specific antigen by differentiating into effector cells, ready to fight an infection. However, unlike naive cells, memory cells do not require the same degree of co-stimulatory signaling for activation, and once activated, they proliferat e and differentiate into effector cells more quickly than naive cells do. \end_layout \begin_layout Standard In the context of a pathogenic infection, immune memory is a major advantage, allowing an organism to rapidly fight off a previously encountered pathogen much more quickly and effectively than the first time it was encountered. However, if effector cells that recognize an antigen from an allograft are allowed to differentiate into memory cells, preventing rejection of the graft becomes much more difficult. Many immune suppression drugs work by interfering with the co-stimulation that naive cells require in order to mount an immune response [CITE?]. Since memory cells do not require this co-stimulation, these drugs are not effective at suppressing an immune response that is mediated by memory cells. Secondly, because memory cells are able to mount a stronger and faster response to an antigen, all else being equal they require stronger immune suppression than naive cells to prevent an immune response. However, immune suppression affects the entire immune system, not just cells recognizing a specific antigen, so increasing the dosage of immune suppression drugs also increases the risk of complications from a compromised immune system, such as opportunistic infections. While the differences in cell surface markers between naive and memory cells have been fairly well characterized, the internal regulatory mechanisms that allow memory cells to respond more quickly and without co-stimulation are still poorly understood. In order to develop methods of immune suppression that either prevent the formation of memory cells or work more effectively against memory cells, the mechanisms of immune memory formation and regulation must be better understood. \end_layout \begin_layout Subsection Overview of bioinformatic analysis methods \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Also cite: R, Bioconductor, snakemake, python, pandas, bedtools, bowtie2, hisat2, STAR, samtools, sra-toolkit, picard tools \end_layout \end_inset \end_layout \begin_layout Standard The studies presented in this work all involve the analysis of high-throughput genomic and epigenomic data. These data present many unique analysis challenges, and a wide array of software tools are available to analyze them. This section presents an overview of the methods used, including what problems they solve, what assumptions they make, and a basic description of how they work. \end_layout \begin_layout Subsubsection \begin_inset Flex Code status open \begin_layout Plain Layout Limma \end_layout \end_inset : The standard linear modeling framework for genomics \end_layout \begin_layout Standard Linear models are a generalization of the \begin_inset Formula $t$ \end_inset -test and ANOVA to arbitrarily complex experimental designs \begin_inset CommandInset citation LatexCommand cite key "chambers:1992" literal "false" \end_inset . In a typical linear model, there is one dependent variable observation per sample and a large number of samples. For example, in a linear model of height as a function of age and sex, there is one height measurement per person. However, when analyzing genomic data, each sample consists of observations of thousands of dependent variables. For example, in an RNA-seq experiment, the dependent variables may be the count of RNA-seq reads for each annotated gene. In abstract terms, each dependent variable being measured is referred to as a feature. The simplest approach to analyzing such data would be to fit the same model independently to each feature. However, this is undesirable for most genomics data sets. Genomics assays like high-throughput sequencing are expensive, and often the process of generating the samples is also quite expensive and time-consumin g. This expense limits the sample sizes typically employed in genomics experiments , and as a result the statistical power of the linear model for each individual feature is likewise limited. However, because thousands of features from the same samples are analyzed together, there is an opportunity to improve the statistical power of the analysis by exploiting shared patterns of variation across features. This is the core feature of \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset , a linear modeling framework designed for genomic data. \begin_inset Flex Code status open \begin_layout Plain Layout Limma \end_layout \end_inset is typically used to analyze expression microarray data, and more recently RNA-seq data, but it can also be used to analyze any other data for which linear modeling is appropriate. \end_layout \begin_layout Standard The central challenge when fitting a linear model is to estimate the variance of the data accurately. Out of all parameters required to evaluate statistical significance of an effect, the variance is the most difficult to estimate when sample sizes are small. A single shared variance could be estimated for all of the features together, and this estimate would be very stable, in contrast to the individual feature variance estimates. However, this would require the assumption that every feature is equally variable, which is known to be false for most genomic data sets. \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset offers a compromise between these two extremes by using a method called empirical Bayes moderation to \begin_inset Quotes eld \end_inset squeeze \begin_inset Quotes erd \end_inset the distribution of estimated variances toward a single common value that represents the variance of an average feature in the data \begin_inset CommandInset citation LatexCommand cite key "Smyth2004" literal "false" \end_inset . While the individual feature variance estimates are not stable, the common variance estimate for the entire data set is quite stable, so using a combinati on of the two yields a variance estimate for each feature with greater precision than the individual feature variances. The trade-off for this improvement is that squeezing each estimated variance toward the common value introduces some bias – the variance will be underestima ted for features with high variance and overestimated for features with low variance. Essentially, \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset assumes that extreme variances are less common than variances close to the common value. The variance estimates from this empirical Bayes procedure are shown empiricall y to yield greater statistical power than either the individual feature variances or the single common value. \end_layout \begin_layout Standard On top of this core framework, \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset also implements many other enhancements that, further relax the assumptions of the model and extend the scope of what kinds of data it can analyze. Instead of squeezing toward a single common variance value, \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset can model the common variance as a function of a covariate, such as average expression \begin_inset CommandInset citation LatexCommand cite key "Law2013" literal "false" \end_inset . This is essential for RNA-seq data, where higher gene counts yield more precise expression measurements and therefore smaller variances than low-count genes. While linear models typically assume that all samples have equal variance, \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset is able to relax this assumption by identifying and down-weighting samples that diverge more strongly from the linear model across many features \begin_inset CommandInset citation LatexCommand cite key "Ritchie2006,Liu2015" literal "false" \end_inset . In addition, \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset is also able to fit simple mixed models incorporating one random effect in addition to the fixed effects represented by an ordinary linear model \begin_inset CommandInset citation LatexCommand cite key "Smyth2005a" literal "false" \end_inset . Once again, \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset shares information between features to obtain a robust estimate for the random effect correlation. \end_layout \begin_layout Subsubsection edgeR provides \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset -like analysis features for count data \end_layout \begin_layout Standard Although \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset can be applied to read counts from RNA-seq data, it is less suitable for counts from ChIP-seq data, which tend to be much smaller and therefore violate the assumption of a normal distribution more severely. For all count-based data, the \begin_inset Flex Code status open \begin_layout Plain Layout edgeR \end_layout \end_inset package works similarly to \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset , but uses a generalized linear model instead of a linear model. The most important difference is that the GLM in \begin_inset Flex Code status open \begin_layout Plain Layout edgeR \end_layout \end_inset models the counts directly using a negative binomial distribution rather than modeling the normalized log counts using a normal distribution \begin_inset CommandInset citation LatexCommand cite key "Chen2014,McCarthy2012,Robinson2010a" literal "false" \end_inset . The negative binomial is a good fit for count data because it can be derived as a gamma-distributed mixture of Poisson distributions. The Poisson distribution accurately represents the distribution of counts expected for a given gene abundance, and the gamma distribution is then used to represent the variation in gene abundance between biological replicates. For this reason, the square root of the dispersion parameter of the negative binomial is sometimes referred to as the biological coefficient of variation, since it represents the variability that was present in the samples prior to the Poisson \begin_inset Quotes eld \end_inset noise \begin_inset Quotes erd \end_inset that was generated by the random sampling of reads in proportion to feature abundances. The choice of a gamma distribution is arbitrary and motivated by mathematical convenience, since a gamma-Poisson mixture yields the numerically tractable negative binomial distribution. Thus, \begin_inset Flex Code status open \begin_layout Plain Layout edgeR \end_layout \end_inset assumes \emph on a prioi \emph default that the variation in abundances between replicates follows a gamma distribution. For differential abundance testing, \begin_inset Flex Code status open \begin_layout Plain Layout edgeR \end_layout \end_inset offers a likelihood ratio test, but more recently recommends a quasi-likelihood test that properly factors the uncertainty in variance estimation into the statistical significance for each feature \begin_inset CommandInset citation LatexCommand cite key "Lund2012" literal "false" \end_inset . \end_layout \begin_layout Subsubsection ChIP-seq Peak calling \end_layout \begin_layout Standard Unlike RNA-seq data, in which gene annotations provide a well-defined set of discrete genomic regions in which to count reads, ChIP-seq reads can potentially occur anywhere in the genome. However, most genome regions will not contain significant ChIP-seq read coverage, and analyzing every position in the entire genome is statistically and computationally infeasible, so it is necessary to identify regions of interest inside which ChIP-seq reads will be counted and analyzed. One option is to define a set of interesting regions \emph on a priori \emph default , for example by defining a promoter region for each annotated gene. However, it is also possible to use the ChIP-seq data itself to identify regions with ChIP-seq read coverage significantly above the background level, known as peaks. \end_layout \begin_layout Standard There are generally two kinds of peaks that can be identified: narrow peaks and broadly enriched regions. Proteins like transcription factors that bind specific sites in the genome typically show most of their ChIP-seq read coverage at these specific sites and very little coverage anywhere else. Because the footprint of the protein is consistent wherever it binds, each peak has a consistent width, typically tens to hundreds of base pairs, representing the length of DNA that it binds to. Algorithms like MACS exploit this pattern to identify specific loci at which such \begin_inset Quotes eld \end_inset narrow peaks \begin_inset Quotes erd \end_inset occur by looking for the characteristic peak shape in the ChIP-seq coverage rising above the surrounding background coverage \begin_inset CommandInset citation LatexCommand cite key "Zhang2008" literal "false" \end_inset . In contrast, some proteins, chief among them histones, do not bind only at a small number of specific sites, but rather bind potentially almost everywhere in the entire genome. When looking at histone marks, adjacent histones tend to be similarly marked, and a given mark may be present on an arbitrary number of consecutive histones along the genome. Hence, there is no consistent \begin_inset Quotes eld \end_inset footprint size \begin_inset Quotes erd \end_inset for ChIP-seq peaks based on histone marks, and peaks typically span many histones. Hence, typical peaks span many hundreds or even thousands of base pairs. Instead of identifying specific loci of strong enrichment, algorithms like SICER assume that peaks are represented in the ChIP-seq data by modest enrichment above background occurring across broad regions, and they attempt to identify the extent of those regions \begin_inset CommandInset citation LatexCommand cite key "Zang2009" literal "false" \end_inset . In all cases, better results are obtained if the local background coverage level can be estimated from ChIP-seq input samples, since various biases can result in uneven background coverage. \end_layout \begin_layout Standard Regardless of the type of peak identified, it is important to identify peaks that occur consistently across biological replicates. The ENCODE project has developed a method called irreproducible discovery rate for this purpose \begin_inset CommandInset citation LatexCommand cite key "Li2006" literal "false" \end_inset . The IDR is defined as the probability that a peak identified in one biological replicate will \emph on not \emph default also be identified in a second replicate. Where the more familiar false discovery rate measures the degree of corresponde nce between a data-derived ranked list and the true list of significant features, IDR instead measures the degree of correspondence between two ranked lists derived from different data. IDR assumes that the highest-ranked features are \begin_inset Quotes eld \end_inset signal \begin_inset Quotes erd \end_inset peaks that tend to be listed in the same order in both lists, while the lowest-ranked features are essentially noise peaks, listed in random order with no correspondence between the lists. IDR attempts to locate the \begin_inset Quotes eld \end_inset crossover point \begin_inset Quotes erd \end_inset between the signal and the noise by determining how far down the list the correspondence between feature ranks breaks down. \end_layout \begin_layout Standard In addition to other considerations, if called peaks are to be used as regions of interest for differential abundance analysis, then care must be taken to call peaks in a way that is blind to differential abundance between experimental conditions, or else the statistical significance calculations for differential abundance will overstate their confidence in the results. The \begin_inset Flex Code status open \begin_layout Plain Layout csaw \end_layout \end_inset package provides guidelines for calling peaks in this way: peaks are called based on a combination of all ChIP-seq reads from all experimental conditions, so that the identified peaks are based on the average abundance across all conditions, which is independent of any differential abundance between conditions \begin_inset CommandInset citation LatexCommand cite key "Lun2015a" literal "false" \end_inset . \end_layout \begin_layout Subsubsection Normalization of high-throughput data is non-trivial and application-dependent \end_layout \begin_layout Standard High-throughput data sets invariably require some kind of normalization before further analysis can be conducted. In general, the goal of normalization is to remove effects in the data that are caused by technical factors that have nothing to do with the biology being studied. \end_layout \begin_layout Standard For Affymetrix expression arrays, the standard normalization algorithm used in most analyses is Robust Multichip Average (RMA) [CITE]. RMA is designed with the assumption that some fraction of probes on each array will be artifactual and takes advantage of the fact that each gene is represented by multiple probes by implementing normalization and summarizati on steps that are robust against outlier probes. However, RMA uses the probe intensities of all arrays in the data set in the normalization of each individual array, meaning that the normalized expression values in each array depend on every array in the data set, and will necessarily change each time an array is added or removed from the data set. If this is undesirable, frozen RMA implements a variant of RMA where the relevant distributional parameters are learned from a large reference set of diverse public array data sets and then \begin_inset Quotes eld \end_inset frozen \begin_inset Quotes erd \end_inset , so that each array is effectively normalized against this frozen reference set rather than the other arrays in the data set under study [CITE]. Other array normalization methods considered include dChip, GRSN, and SCAN [CITEx3]. \end_layout \begin_layout Standard In contrast, high-throughput sequencing data present very different normalizatio n challenges. The simplest case is RNA-seq in which read counts are obtained for a set of gene annotations, yielding a matrix of counts with rows representing genes and columns representing samples. Because RNA-seq approximates a process of sampling from a population with replacement, each gene's count is only interpretable as a fraction of the total reads for that sample. For that reason, RNA-seq abundances are often reported as counts per million (CPM). Furthermore, if the abundance of a single gene increases, then in order for its fraction of the total reads to increase, all other genes' fractions must decrease to accommodate it. This effect is known as composition bias, and it is an artifact of the read sampling process that has nothing to do with the biology of the samples and must therefore be normalized out. The most commonly used methods to normalize for composition bias in RNA-seq data seek to equalize the average gene abundance across samples, under the assumption that the average gene is likely not changing \begin_inset CommandInset citation LatexCommand cite key "Robinson2010,Anders2010" literal "false" \end_inset . \end_layout \begin_layout Standard In ChIP-seq data, normalization is not as straightforward. The \begin_inset Flex Code status open \begin_layout Plain Layout csaw \end_layout \end_inset package implements several different normalization strategies and provides guidance on when to use each one \begin_inset CommandInset citation LatexCommand cite key "Lun2015a" literal "false" \end_inset . Briefly, a typical ChIP-seq sample has a bimodal distribution of read counts: a low-abundance mode representing background regions and a high-abundance mode representing signal regions. This offers two potential normalization targets: equalizing background coverage or equalizing signal coverage. If the experiment is well controlled and ChIP efficiency is known to be consistent across all samples, then normalizing the background coverage to be equal across all samples is a reasonable strategy. If this is not a safe assumption, then the preferred strategy is to normalize the signal regions in a way similar to RNA-seq data by assuming that the average signal region is not changing abundance between samples. Beyond this, if a ChIP-seq experiment has a more complicated structure that doesn't show the typical bimodal count distribution, it may be necessary to implement a normalization as a smooth function of abundance. However, this strategy makes a much stronger assumption about the data: that the average log fold change is zero across all abundance levels. Hence, the simpler scaling normalization based on background or signal regions are generally preferred whenever possible. \end_layout \begin_layout Subsubsection ComBat and SVA for correction of known and unknown batch effects \end_layout \begin_layout Standard In addition to well-understood effects that can be easily normalized out, a data set often contains confounding biological effects that must be accounted for in the modeling step. For instance, in an experiment with pre-treatment and post-treatment samples of cells from several different donors, donor variability represents a known batch effect. The most straightforward correction for known batches is to estimate the mean for each batch independently and subtract out the differences, so that all batches have identical means for each feature. However, as with variance estimation, estimating the differences in batch means is not necessarily robust at the feature level, so the ComBat method adds empirical Bayes squeezing of the batch mean differences toward a common value, analogous to \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset 's empirical Bayes squeezing of feature variance estimates \begin_inset CommandInset citation LatexCommand cite key "Johnson2007" literal "false" \end_inset . Effectively, ComBat assumes that modest differences between batch means are real batch effects, but extreme differences between batch means are more likely to be the result of outlier observations that happen to line up with the batches rather than a genuine batch effect. The result is a batch correction that is more robust against outliers than simple subtraction of mean differences subtraction. \end_layout \begin_layout Standard In some data sets, unknown batch effects may be present due to inherent variability in in the data, either caused by technical or biological effects. Examples of unknown batch effects include variations in enrichment efficiency between ChIP-seq samples, variations in populations of different cell types, and the effects of uncontrolled environmental factors on gene expression in humans or live animals. In an ordinary linear model context, unknown batch effects cannot be inferred and must be treated as random noise. However, in high-throughput experiments, once again information can be shared across features to identify patterns of un-modeled variation that are repeated in many features. One attractive strategy would be to perform singular value decomposition (SVD) on the matrix of linear model residuals (which contain all the un-modeled variation in the data) and take the first few singular vectors as batch effects. While this can be effective, it makes the unreasonable assumption that all batch effects are uncorrelated with any of the effects being modeled. Surrogate variable analysis (SVA) starts with this approach, but takes some additional steps to identify batch effects in the full data that are both highly correlated with the singular vectors in the residuals and least correlated with the effects of interest \begin_inset CommandInset citation LatexCommand cite key "Leek2007" literal "false" \end_inset . Since the final batch effects are estimated from the full data, moderate correlations between the batch effects and effects of interest are allowed, which gives SVA much more freedom to estimate the true extent of the batch effects compared to simple residual SVD. Once the surrogate variables are estimated, they can be included as coefficient s in the linear model in a similar fashion to known batch effects in order to subtract out their effects on each feature's abundance. \end_layout \begin_layout Subsubsection Factor analysis: PCA, MDS, MOFA \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Not sure if this merits a subsection here. \end_layout \end_inset \end_layout \begin_layout Itemize Batch-corrected PCA is informative, but careful application is required to avoid bias \end_layout \begin_layout Section Innovation \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Is this entire section redundant with the Approach sections of each chapter? I'm not really sure what to write here. \end_layout \end_inset \end_layout \begin_layout Subsection MSC infusion to improve transplant outcomes (prevent/delay rejection) \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Do I still talk about this? It's the motivation for chapter 4, but I don't actually present any work related to MSCs. \end_layout \end_inset \end_layout \begin_layout Itemize Demonstrated in mice, but not yet in primates \end_layout \begin_layout Itemize Mechanism currently unknown, but MSC are known to be immune modulatory \end_layout \begin_layout Itemize Characterize MSC response to interferon gamma \end_layout \begin_layout Itemize IFN-g is thought to stimulate their function \end_layout \begin_layout Itemize Test IFN-g treated MSC infusion as a therapy to delay graft rejection in cynomolgus monkeys \end_layout \begin_layout Itemize Monitor animals post-transplant using blood RNA-seq at serial time points \end_layout \begin_layout Subsection Investigate dynamics of histone marks in CD4 T-cell activation and memory \end_layout \begin_layout Itemize Previous studies have looked at single snapshots of histone marks \end_layout \begin_layout Itemize Instead, look at changes in histone marks across activation and memory \end_layout \begin_layout Subsection High-throughput sequencing and microarray technologies \end_layout \begin_layout Itemize Powerful methods for assaying gene expression and epigenetics across entire genomes \end_layout \begin_layout Itemize Proper analysis requires finding and exploiting systematic genome-wide trends \end_layout \begin_layout Chapter Reproducible genome-wide epigenetic analysis of H3K4 and H3K27 methylation in naive and memory CD4 T-cell activation \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Chapter author list: Me, Sarah, Dan \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Need better section titles throughout the entire chapter \end_layout \end_inset \end_layout \begin_layout Section Approach \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Check on the exact correct way to write \begin_inset Quotes eld \end_inset CD4 T-cell \begin_inset Quotes erd \end_inset . I think there might be a plus sign somewhere in there now? Also, maybe figure out a reasonable way to abbreviate \begin_inset Quotes eld \end_inset naive CD4 T-cells \begin_inset Quotes erd \end_inset and \begin_inset Quotes eld \end_inset memory CD4 T-cells \begin_inset Quotes erd \end_inset . \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Is it ok to just copy a bunch of citations from the intros to Sarah's papers? That feels like cheating somehow. \end_layout \end_inset \end_layout \begin_layout Standard CD4 T-cells are central to all adaptive immune responses, as well as immune memory [CITE?]. After an infection is cleared, a subset of the naive CD4 T-cells that responded to that infection differentiate into memory CD4 T-cells, which are responsible for responding to the same pathogen in the future. Memory CD4 T-cells are functionally distinct, able to respond to an infection more quickly and without the co-stimulation required by naive CD4 T-cells. However, the molecular mechanisms underlying this functional distinction are not well-understood. Epigenetic regulation via histone modification is thought to play an important role, but while many studies have looked at static snapshots of histone methylation in T-cells, few studies have looked at the dynamics of histone regulation after T-cell activation, nor the differences in histone methylation between naive and memory T-cells. H3K4me2, H3K4me3 and H3K27me3 are three histone marks thought to be major epigenetic regulators of gene expression. The goal of the present study is to investigate the role of these histone marks in CD4 T-cell activation kinetics and memory differentiation. In static snapshots, H3K4me2 and H3K4me3 are often observed in the promoters of highly transcribed genes, while H3K27me3 is more often observed in promoters of inactive genes with little to no transcription occurring. As a result, the two H3K4 marks have been characterized as \begin_inset Quotes eld \end_inset activating \begin_inset Quotes erd \end_inset marks, while H3K27me3 has been characterized as \begin_inset Quotes eld \end_inset deactivating \begin_inset Quotes erd \end_inset . Despite these characterizations, the actual causal relationship between these histone modifications and gene transcription is complex and likely involves positive and negative feedback loops between the two. \end_layout \begin_layout Standard In order to investigate the relationship between gene expression and these histone modifications in the context of naive and memory CD4 T-cell activation, a previously published data set of combined RNA-seq and ChIP-seq data was re-analyzed using up-to-date methods designed to address the specific analysis challenges posed by this data set. The data set contains naive and memory CD4 T-cell samples in a time course before and after activation. Like the original analysis, this analysis looks at the dynamics of these marks histone marks and compare them to gene expression dynamics at the same time points during activation, as well as compare them between naive and memory cells, in hope of discovering evidence of new mechanistic details in the interplay between them. The original analysis of this data treated each gene promoter as a monolithic unit and mostly assumed that ChIP-seq reads or peaks occurring anywhere within a promoter were equivalent, regardless of where they occurred relative to the gene structure. For an initial analysis of the data, this was a necessary simplifying assumptio n. The current analysis aims to relax this assumption, first by directly analyzing ChIP-seq peaks for differential modification, and second by taking a more granular look at the ChIP-seq read coverage within promoter regions to ask whether the location of histone modifications relative to the gene's TSS is an important factor, as opposed to simple proximity. \end_layout \begin_layout Section Methods \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Look up some more details from the papers (e.g. activation method). \end_layout \end_inset \end_layout \begin_layout Standard A reproducible workflow was written to analyze the raw ChIP-seq and RNA-seq data from previous studies \begin_inset CommandInset citation LatexCommand cite key "gh-cd4-csaw,LaMere2016,LaMere2017" literal "true" \end_inset . Briefly, this data consists of RNA-seq and ChIP-seq from CD4 T-cells cultured from 4 donors. From each donor, naive and memory CD4 T-cells were isolated separately. Then cultures of both cells were activated [how?], and samples were taken at 4 time points: Day 0 (pre-activation), Day 1 (early activation), Day 5 (peak activation), and Day 14 (post-activation). For each combination of cell type and time point, RNA was isolated and sequenced, and ChIP-seq was performed for each of 3 histone marks: H3K4me2, H3K4me3, and H3K27me3. The ChIP-seq input DNA was also sequenced for each sample. The result was 32 samples for each assay. \end_layout \begin_layout Subsection RNA-seq differential expression analysis \end_layout \begin_layout Standard \begin_inset Note Note status collapsed \begin_layout Plain Layout \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/rnaseq-compare/ensmebl-vs-entrez-star-CROP.png lyxscale 25 width 35col% groupId rna-comp-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout STAR quantification, Entrez vs Ensembl gene annotation \end_layout \end_inset \end_layout \end_inset \begin_inset space \qquad{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/rnaseq-compare/ensmebl-vs-entrez-shoal-CROP.png lyxscale 25 width 35col% groupId rna-comp-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout Salmon+Shoal quantification, Entrez vs Ensembl gene annotation \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/rnaseq-compare/star-vs-hisat2-CROP.png lyxscale 25 width 35col% groupId rna-comp-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout STAR vs HISAT2 quantification, Ensembl gene annotation \end_layout \end_inset \end_layout \end_inset \begin_inset space \qquad{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/rnaseq-compare/star-vs-salmon-CROP.png lyxscale 25 width 35col% groupId rna-comp-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout Salmon vs STAR quantification, Ensembl gene annotation \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/rnaseq-compare/salmon-vs-kallisto-CROP.png lyxscale 25 width 35col% groupId rna-comp-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout Salmon vs Kallisto quantification, Ensembl gene annotation \end_layout \end_inset \end_layout \end_inset \begin_inset space \qquad{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/rnaseq-compare/salmon-vs-shoal-CROP.png lyxscale 25 width 35col% groupId rna-comp-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout Salmon+Shoal vs Salmon alone, Ensembl gene annotation \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:RNA-norm-comp" \end_inset RNA-seq comparisons \end_layout \end_inset \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard Sequence reads were retrieved from the Sequence Read Archive (SRA) \begin_inset CommandInset citation LatexCommand cite key "Leinonen2011" literal "false" \end_inset . Five different alignment and quantification methods were tested for the RNA-seq data \begin_inset CommandInset citation LatexCommand cite key "Dobin2012,Kim2019,Liao2014,Pimentel2016,Patro2017,gh-shoal,gh-hg38-ref" literal "false" \end_inset . Each quantification was tested with both Ensembl transcripts and UCSC known gene annotations [CITE? Also which versions of each?]. Comparisons of downstream results from each combination of quantification method and reference revealed that all quantifications gave broadly similar results for most genes, so shoal with the Ensembl annotation was chosen as the method theoretically most likely to partially mitigate some of the batch effect in the data. \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/RNA-seq/PCA-no-batchsub-CROP.png lyxscale 25 width 75col% groupId rna-pca-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:RNA-PCA-no-batchsub" \end_inset Before batch correction \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/RNA-seq/PCA-combat-batchsub-CROP.png lyxscale 25 width 75col% groupId rna-pca-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:RNA-PCA-ComBat-batchsub" \end_inset After batch correction with ComBat \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:RNA-PCA" \end_inset PCoA plots of RNA-seq data showing effect of batch correction. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard Due to an error in sample preparation, the RNA from the samples for days 0 and 5 were sequenced using a different kit than those for days 1 and 14. This induced a substantial batch effect in the data due to differences in sequencing biases between the two kits, and this batch effect is unfortunate ly confounded with the time point variable (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:RNA-PCA-no-batchsub" plural "false" caps "false" noprefix "false" \end_inset ). To do the best possible analysis with this data, this batch effect was subtracted out from the data using ComBat \begin_inset CommandInset citation LatexCommand cite key "Johnson2007" literal "false" \end_inset , ignoring the time point variable due to the confounding with the batch variable. The result is a marked improvement, but the unavoidable confounding with time point means that certain real patterns of gene expression will be indistinguishable from the batch effect and subtracted out as a result. Specifically, any \begin_inset Quotes eld \end_inset zig-zag \begin_inset Quotes erd \end_inset pattern, such as a gene whose expression goes up on day 1, down on day 5, and back up again on day 14, will be attenuated or eliminated entirely. In the context of a T-cell activation time course, it is unlikely that many genes of interest will follow such an expression pattern, so this loss was deemed an acceptable cost for correcting the batch effect. \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Just take the top row \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/RNA-seq/weights-vs-covars-CROP.png lyxscale 25 width 100col% groupId colwidth-raster \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:RNA-seq-weights-vs-covars" \end_inset RNA-seq sample weights, grouped by experimental and technical covariates. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard However, removing the systematic component of the batch effect still leaves the noise component. The gene quantifications from the first batch are substantially noisier than those in the second batch. This analysis corrected for this by using \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset 's sample weighting method to assign lower weights to the noisy samples of batch 1 \begin_inset CommandInset citation LatexCommand cite key "Ritchie2006,Liu2015" literal "false" \end_inset . The resulting analysis gives an accurate assessment of statistical significance for all comparisons, which unfortunately means a loss of statistical power for comparisons involving samples in batch 1. \end_layout \begin_layout Standard In any case, the RNA-seq counts were first normalized using trimmed mean of M-values \begin_inset CommandInset citation LatexCommand cite key "Robinson2010" literal "false" \end_inset , converted to normalized logCPM with quality weights using \begin_inset Flex Code status open \begin_layout Plain Layout voomWithQualityWeights \end_layout \end_inset \begin_inset CommandInset citation LatexCommand cite key "Law2013,Liu2015" literal "false" \end_inset , and batch-corrected at this point using ComBat. A linear model was fit to the batch-corrected, quality-weighted data for each gene using \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset , and each gene was tested for differential expression using \begin_inset Flex Code status open \begin_layout Plain Layout limma \end_layout \end_inset 's empirical Bayes moderated \begin_inset Formula $t$ \end_inset -test \begin_inset CommandInset citation LatexCommand cite key "Smyth2005,Law2013,Phipson2013" literal "false" \end_inset . \end_layout \begin_layout Subsection ChIP-seq differential modification analysis \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/csaw/CCF-plots-noBL-PAGE2-CROP.pdf lyxscale 50 height 40theight% groupId ccf-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:CCF-without-blacklist" \end_inset Cross-correlation plots without removing blacklisted reads. \series default Without blacklisting, many artifactual peaks are visible in the cross-correlatio ns of the ChIP-seq samples, and the peak at the true fragment size (147 \begin_inset space ~ \end_inset bp) is frequently overshadowed by the artifactual peak at the read length (100 \begin_inset space ~ \end_inset bp). \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/csaw/CCF-plots-PAGE2-CROP.pdf lyxscale 50 height 40theight% groupId ccf-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:CCF-with-blacklist" \end_inset Cross-correlation plots with blacklisted reads removed. \series default After blacklisting, most ChIP-seq samples have clean-looking periodic cross-cor relation plots, with the largest peak around 147 \begin_inset space ~ \end_inset bp, the expected size for a fragment of DNA from a single nucleosome, and little to no peak at the read length, 100 \begin_inset space ~ \end_inset bp. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:CCF-master" \end_inset Strand cross-correlation plots for ChIP-seq data, before and after blacklisting. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Note Note status open \begin_layout Plain Layout \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K4me2-sample-MAplot-bins-CROP.png lyxscale 25 width 100col% groupId colwidth-raster \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:MA-plot-bigbins" \end_inset MA plot of H3K4me2 read counts in 10kb bins for two arbitrary samples. \end_layout \end_inset \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Be consistent about use of \begin_inset Quotes eld \end_inset differential binding \begin_inset Quotes erd \end_inset vs \begin_inset Quotes eld \end_inset differential modification \begin_inset Quotes erd \end_inset throughout this chapter. The latter is usually preferred. \end_layout \end_inset \end_layout \begin_layout Standard Sequence reads were retrieved from SRA \begin_inset CommandInset citation LatexCommand cite key "Leinonen2011" literal "false" \end_inset . ChIP-seq (and input) reads were aligned to GRCh38 genome assembly using Bowtie 2 \begin_inset CommandInset citation LatexCommand cite key "Langmead2012,Schneider2017,gh-hg38-ref" literal "false" \end_inset . Artifact regions were annotated using a custom implementation of the GreyListCh IP algorithm, and these \begin_inset Quotes eld \end_inset greylists \begin_inset Quotes erd \end_inset were merged with the published ENCODE blacklists \begin_inset CommandInset citation LatexCommand cite key "greylistchip,Amemiya2019,Dunham2012,gh-cd4-csaw" literal "false" \end_inset . Any read or called peak overlapping one of these regions was regarded as artifactual and excluded from downstream analyses. Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:CCF-master" plural "false" caps "false" noprefix "false" \end_inset shows the improvement after blacklisting in the strand cross-correlation plots, a common quality control plot for ChIP-seq data. Peaks were called using epic, an implementation of the SICER algorithm \begin_inset CommandInset citation LatexCommand cite key "Zang2009,gh-epic" literal "false" \end_inset . Peaks were also called separately using MACS, but MACS was determined to be a poor fit for the data, and these peak calls are not used in any further analyses \begin_inset CommandInset citation LatexCommand cite key "Zhang2008" literal "false" \end_inset . Consensus peaks were determined by applying the irreproducible discovery rate (IDR) framework \begin_inset CommandInset citation LatexCommand cite key "Li2006,gh-idr" literal "false" \end_inset to find peaks consistently called in the same locations across all 4 donors. \end_layout \begin_layout Standard Promoters were defined by computing the distance from each annotated TSS to the nearest called peak and examining the distribution of distances, observing that peaks for each histone mark were enriched within a certain distance of the TSS. For H3K4me2 and H3K4me3, this distance was about 1 \begin_inset space ~ \end_inset kb, while for H3K27me3 it was 2.5 \begin_inset space ~ \end_inset kb. These distances were used as an \begin_inset Quotes eld \end_inset effective promoter radius \begin_inset Quotes erd \end_inset for each mark. The promoter region for each gene was defined as the region of the genome within this distance upstream or downstream of the gene's annotated TSS. For genes with multiple annotated TSSs, a promoter region was defined for each TSS individually, and any promoters that overlapped (due to multiple TSSs being closer than 2 times the radius) were merged into one large promoter. Thus, some genes had multiple promoters defined, which were each analyzed separately for differential modification. \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K4me2-PCA-raw-CROP.png lyxscale 25 width 45col% groupId pcoa-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:PCoA-H3K4me2-bad" \end_inset H3K4me2, no correction \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K4me2-PCA-SVsub-CROP.png lyxscale 25 width 45col% groupId pcoa-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:PCoA-H3K4me2-good" \end_inset H3K4me2, SVs subtracted \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K4me3-PCA-raw-CROP.png lyxscale 25 width 45col% groupId pcoa-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:PCoA-H3K4me3-bad" \end_inset H3K4me3, no correction \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K4me3-PCA-SVsub-CROP.png lyxscale 25 width 45col% groupId pcoa-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:PCoA-H3K4me3-good" \end_inset H3K4me3, SVs subtracted \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K27me3-PCA-raw-CROP.png lyxscale 25 width 45col% groupId pcoa-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:PCoA-H3K27me3-bad" \end_inset H3K27me3, no correction \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K27me3-PCA-SVsub-CROP.png lyxscale 25 width 45col% groupId pcoa-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:PCoA-H3K27me3-good" \end_inset H3K27me3, SVs subtracted \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:PCoA-ChIP" \end_inset PCoA plots of ChIP-seq sliding window data, before and after subtracting surrogate variables (SVs). \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard Reads in promoters, peaks, and sliding windows across the genome were counted and normalized using \begin_inset Flex Code status open \begin_layout Plain Layout csaw \end_layout \end_inset and analyzed for differential modification using \begin_inset Flex Code status open \begin_layout Plain Layout edgeR \end_layout \end_inset \begin_inset CommandInset citation LatexCommand cite key "Lun2014,Lun2015a,Lund2012,Phipson2016" literal "false" \end_inset . Unobserved confounding factors in the ChIP-seq data were corrected using SVA \begin_inset CommandInset citation LatexCommand cite key "Leek2007,Leek2014" literal "false" \end_inset . Principal coordinate plots of the promoter count data for each histone mark before and after subtracting surrogate variable effects are shown in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:PCoA-ChIP" plural "false" caps "false" noprefix "false" \end_inset . \end_layout \begin_layout Standard To investigate whether the location of a peak within the promoter region was important, \begin_inset Quotes eld \end_inset relative coverage profiles \begin_inset Quotes erd \end_inset were generated. First, 500-bp sliding windows were tiled around each annotated TSS: one window centered on the TSS itself, and 10 windows each upstream and downstream, thus covering a 10.5-kb region centered on the TSS with 21 windows. Reads in each window for each TSS were counted in each sample, and the counts were normalized and converted to log CPM as in the differential modification analysis. Then, the logCPM values within each promoter were normalized to an average of zero, such that each window's normalized abundance now represents the relative read depth of that window compared to all other windows in the same promoter. The normalized abundance values for each window in a promoter are collectively referred to as that promoter's \begin_inset Quotes eld \end_inset relative coverage profile \begin_inset Quotes erd \end_inset . \end_layout \begin_layout Subsection MOFA recovers biologically relevant variation from blind analysis by correlating across datasets \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash afterpage{ \end_layout \begin_layout Plain Layout \backslash begin{landscape} \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/MOFA-varExplaiend-matrix-CROP.png lyxscale 25 width 45col% groupId mofa-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:mofa-varexplained" \end_inset Variance explained in each data set by each latent factor estimated by MOFA. \series default For each latent factor (LF) learned by MOFA, the variance explained by that factor in each data set ( \begin_inset Quotes eld \end_inset view \begin_inset Quotes erd \end_inset ) is shown by the shading of the cells in the lower section. The upper section shows the total fraction of each data set's variance that is explained by all LFs combined. \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/MOFA-LF-scatter-CROP.png lyxscale 25 width 45col% groupId mofa-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:mofa-lf-scatter" \end_inset Scatter plots of specific pairs of MOFA latent factors. \series default LFs 1, 4, and 5 explain substantial variation in all data sets, so they are plotted against each other in order to reveal patterns of variation that are shared across all data sets. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:MOFA-master" \end_inset MOFA latent factors separate technical confounders from \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash end{landscape} \end_layout \begin_layout Plain Layout } \end_layout \end_inset \end_layout \begin_layout Standard MOFA was run on all the ChIP-seq windows overlapping consensus peaks for each histone mark, as well as the RNA-seq data, in order to identify patterns of coordinated variation across all data sets \begin_inset CommandInset citation LatexCommand cite key "Argelaguet2018" literal "false" \end_inset . The results are summarized in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:MOFA-master" plural "false" caps "false" noprefix "false" \end_inset . Latent factors 1, 4, and 5 were determined to explain the most variation consistently across all data sets (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:mofa-varexplained" plural "false" caps "false" noprefix "false" \end_inset ), and scatter plots of these factors show that they also correlate best with the experimental factors (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:mofa-lf-scatter" plural "false" caps "false" noprefix "false" \end_inset ). Latent factor 2 captures the batch effect in the RNA-seq data. Removing the effect of LF2 using MOFA theoretically yields a batch correction that does not depend on knowing the experimental factors. When this was attempted, the resulting batch correction was comparable to ComBat (see Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:RNA-PCA-ComBat-batchsub" plural "false" caps "false" noprefix "false" \end_inset ), indicating that the ComBat-based batch correction has little room for improvement given the problems with the data set. \end_layout \begin_layout Standard \begin_inset Note Note status collapsed \begin_layout Plain Layout \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/MOFA-batch-correct-CROP.png lyxscale 25 width 100col% groupId colwidth-raster \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:mofa-batchsub" \end_inset Result of RNA-seq batch-correction using MOFA latent factors \end_layout \end_inset \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Section Results \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Focus on what hypotheses were tested, then select figures that show how those hypotheses were tested, even if the result is a negative. Not every interesting result needs to be in here. Chapter should tell a story. \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Maybe reorder these sections to do RNA-seq, then ChIP-seq, then combined analyses? \end_layout \end_inset \end_layout \begin_layout Subsection Interpretation of RNA-seq analysis is limited by a major confounding factor \end_layout \begin_layout Standard \begin_inset Float table wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Tabular \begin_inset Text \begin_layout Plain Layout Test \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Est. non-null \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \begin_inset Formula $\mathrm{FDR}\le10\%$ \end_inset \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Naive Day 0 vs Day 1 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 5992 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 1613 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Naive Day 0 vs Day 5 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 3038 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 32 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Naive Day 0 vs Day 14 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 1870 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 190 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Memory Day 0 vs Day 1 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 3195 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 411 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Memory Day 0 vs Day 5 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 2688 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 18 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Memory Day 0 vs Day 14 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 1911 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 227 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Day 0 Naive vs Memory \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 2 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Day 1 Naive vs Memory \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 9167 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 5532 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Day 5 Naive vs Memory \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Day 14 Naive vs Memory \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 6446 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 2319 \end_layout \end_inset \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "tab:Estimated-and-detected-rnaseq" \end_inset Estimated and detected differentially expressed genes. \series default \begin_inset Quotes eld \end_inset Test \begin_inset Quotes erd \end_inset : Which sample groups were compared; \begin_inset Quotes eld \end_inset Est non-null \begin_inset Quotes erd \end_inset : Estimated number of differentially expressed genes, using the method of averaging local FDR values \begin_inset CommandInset citation LatexCommand cite key "Phipson2013Thesis" literal "false" \end_inset ; \begin_inset Quotes eld \end_inset \begin_inset Formula $\mathrm{FDR}\le10\%$ \end_inset \begin_inset Quotes erd \end_inset : Number of significantly differentially expressed genes at an FDR threshold of 10%. The total number of genes tested was 16707. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/RNA-seq/PCA-final-12-CROP.png lyxscale 25 width 100col% groupId colwidth-raster \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:rna-pca-final" \end_inset PCoA plot of RNA-seq samples after ComBat batch correction. \series default Each point represents an individual sample. Samples with the same combination of cell type and time point are encircled with a shaded region to aid in visual identification of the sample groups. Samples with of same cell type from the same donor are connected by lines to indicate the \begin_inset Quotes eld \end_inset trajectory \begin_inset Quotes erd \end_inset of each donor's cells over time in PCoA space. \end_layout \end_inset \end_layout \begin_layout Plain Layout \end_layout \end_inset \end_layout \begin_layout Standard Genes called present in the RNA-seq data were tested for differential expression between all time points and cell types. The counts of differentially expressed genes are shown in Table \begin_inset CommandInset ref LatexCommand ref reference "tab:Estimated-and-detected-rnaseq" plural "false" caps "false" noprefix "false" \end_inset . Notably, all the results for Day 0 and Day 5 have substantially fewer genes called differentially expressed than any of the results for other time points. This is an unfortunate result of the difference in sample quality between the two batches of RNA-seq data. All the samples in Batch 1, which includes all the samples from Days 0 and 5, have substantially more variability than the samples in Batch 2, which includes the other time points. This is reflected in the substantially higher weights assigned to Batch 2 (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:RNA-seq-weights-vs-covars" plural "false" caps "false" noprefix "false" \end_inset ). The batch effect has both a systematic component and a random noise component. While the systematic component was subtracted out using ComBat (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:RNA-PCA" plural "false" caps "false" noprefix "false" \end_inset ), no such correction is possible for the noise component: Batch 1 simply has substantially more random noise in it, which reduces the statistical power for any differential expression tests involving samples in that batch. \end_layout \begin_layout Standard Despite the difficulty in detecting specific differentially expressed genes, there is still evidence that differential expression is present for these time points. In Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:rna-pca-final" plural "false" caps "false" noprefix "false" \end_inset , there is a clear separation between naive and memory samples at Day 0, despite the fact that only 2 genes were significantly differentially expressed for this comparison. Similarly, the small numbers of genes detected for the Day 0 vs Day 5 compariso ns do not reflect the large separation between these time points in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:rna-pca-final" plural "false" caps "false" noprefix "false" \end_inset . In addition, the MOFA latent factor plots in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:mofa-lf-scatter" plural "false" caps "false" noprefix "false" \end_inset . This suggests that there is indeed a differential expression signal present in the data for these comparisons, but the large variability in the Batch 1 samples obfuscates this signal at the individual gene level. As a result, it is impossible to make any meaningful statements about the \begin_inset Quotes eld \end_inset size \begin_inset Quotes erd \end_inset of the gene signature for any time point, since the number of significant genes as well as the estimated number of differentially expressed genes depends so strongly on the variations in sample quality in addition to the size of the differential expression signal in the data. Gene-set enrichment analyses are similarly impractical. However, analyses looking at genome-wide patterns of expression are still practical. \end_layout \begin_layout Subsection H3K4 and H3K27 methylation occur in broad regions and are enriched near promoters \end_layout \begin_layout Standard \begin_inset Float table wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Also get \emph on median \emph default peak width and maybe other quantiles (25%, 75%) \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \begin_inset Tabular \begin_inset Text \begin_layout Plain Layout Histone Mark \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout # Peaks \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Mean peak width \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout genome coverage \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout FRiP \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout H3K4me2 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 14965 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 3970 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 1.92% \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 14.2% \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout H3K4me3 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 6163 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 2946 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0.588% \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 6.57% \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout H3K27me3 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 18139 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 18967 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 11.1% \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 22.5% \end_layout \end_inset \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "tab:peak-calling-summary" \end_inset Peak-calling summary. \series default For each histone mark, the number of peaks called using SICER at an IDR threshold of ???, the mean width of those peaks, the fraction of the genome covered by peaks, and the fraction of reads in peaks (FRiP). \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard Table \begin_inset CommandInset ref LatexCommand ref reference "tab:peak-calling-summary" plural "false" caps "false" noprefix "false" \end_inset gives a summary of the peak calling statistics for each histone mark. Consistent with previous observations [CITATION NEEDED], all 3 histone marks occur in broad regions spanning many consecutive nucleosomes, rather than in sharp peaks as would be expected for a transcription factor or other molecule that binds to specific sites. This conclusion is further supported by Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:CCF-with-blacklist" plural "false" caps "false" noprefix "false" \end_inset , in which a clear nucleosome-sized periodicity is visible in the cross-correlat ion value for each sample, indicating that each time a given mark is present on one histone, it is also likely to be found on adjacent histones as well. H3K27me3 enrichment in particular is substantially more broad than either H3K4 mark, with a mean peak width of almost 19,000 bp. This is also reflected in the periodicity observed in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:CCF-with-blacklist" plural "false" caps "false" noprefix "false" \end_inset , which remains strong much farther out for H3K27me3 than the other marks, showing H3K27me3 especially tends to be found on long runs of consecutive histones. \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Ensure this figure uses the peak calls from the new analysis. \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Need a control: shuffle all peaks and repeat, N times. Do real vs shuffled control both in a top/bottom arrangement. \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Consider counting TSS inside peaks as negative number indicating how far \emph on inside \emph default the peak the TSS is (i.e. distance to nearest non-peak area). \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout The H3K4 part of this figure is included in \begin_inset CommandInset citation LatexCommand cite key "LaMere2016" literal "false" \end_inset as Fig. S2. Do I need to do anything about that? \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/Promoter Peak Distance Profile-PAGE1-CROP.pdf lyxscale 50 width 80col% \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:near-promoter-peak-enrich" \end_inset Enrichment of peaks in promoter neighborhoods. \series default This plot shows the distribution of distances from each annotated transcription start site in the genome to the nearest called peak. Each line represents one combination of histone mark, cell type, and time point. Distributions are smoothed using kernel density estimation [CITE? see ggplot2 stat_density()]. Transcription start sites that occur \emph on within \emph default peaks were excluded from this plot to avoid a large spike at zero that would overshadow the rest of the distribution. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float table wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Tabular \begin_inset Text \begin_layout Plain Layout Histone mark \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Effective promoter radius \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout H3K4me2 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 1 kb \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout H3K4me3 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 1 kb \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout H3K27me3 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 2.5 kb \end_layout \end_inset \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "tab:effective-promoter-radius" \end_inset Effective promoter radius for each histone mark. \series default These values represent the approximate distance from transcription start site positions within which an excess of peaks are found, as shown in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:near-promoter-peak-enrich" plural "false" caps "false" noprefix "false" \end_inset . \end_layout \end_inset \end_layout \begin_layout Plain Layout \end_layout \end_inset \end_layout \begin_layout Standard All 3 histone marks tend to occur more often near promoter regions, as shown in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:near-promoter-peak-enrich" plural "false" caps "false" noprefix "false" \end_inset . The majority of each density distribution is flat, representing the background density of peaks genome-wide. Each distribution has a peak near zero, representing an enrichment of peaks close transcription start site (TSS) positions relative to the remainder of the genome. Interestingly, the \begin_inset Quotes eld \end_inset radius \begin_inset Quotes erd \end_inset within which this enrichment occurs is not the same for every histone mark (Table \begin_inset CommandInset ref LatexCommand ref reference "tab:effective-promoter-radius" plural "false" caps "false" noprefix "false" \end_inset ). For H3K4me2 and H3K4me3, peaks are most enriched within 1 \begin_inset space ~ \end_inset kbp of TSS positions, while for H3K27me3, enrichment is broader, extending to 2.5 \begin_inset space ~ \end_inset kbp. These \begin_inset Quotes eld \end_inset effective promoter radii \begin_inset Quotes erd \end_inset remain approximately the same across all combinations of experimental condition (cell type, time point, and donor), so they appear to be a property of the histone mark itself. Hence, these radii were used to define the promoter regions for each histone mark in all further analyses. \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Consider also showing figure for distance to nearest peak center, and reference median peak size once that is known. \end_layout \end_inset \end_layout \begin_layout Subsection H3K4 and H3K27 promoter methylation has broadly the expected correlation with gene expression \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout This figure is generated from the old analysis. Either note that in some way or re-generate it from the new peak calls. \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/FPKM by Peak Violin Plots-CROP.pdf lyxscale 50 width 100col% \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:fpkm-by-peak" \end_inset Expression distributions of genes with and without promoter peaks. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard H3K4me2 and H3K4me2 have previously been reported as activating marks whose presence in a gene's promoter is associated with higher gene expression, while H3K27me3 has been reported as inactivating [CITE]. The data are consistent with this characterization: genes whose promoters (as defined by the radii for each histone mark listed in \begin_inset CommandInset ref LatexCommand ref reference "tab:effective-promoter-radius" plural "false" caps "false" noprefix "false" \end_inset ) overlap with a H3K4me2 or H3K4me3 peak tend to have higher expression than those that don't, while H3K27me3 is likewise associated with lower gene expression, as shown in \begin_inset CommandInset ref LatexCommand ref reference "fig:fpkm-by-peak" plural "false" caps "false" noprefix "false" \end_inset . This pattern holds across all combinations of cell type and time point (Welch's \emph on t \emph default -test, all \begin_inset Formula $p\mathrm{-values}\ll2.2\times10^{-16}$ \end_inset ). The difference in average log FPKM values when a peak overlaps the promoter is about \begin_inset Formula $+5.67$ \end_inset for H3K4me2, \begin_inset Formula $+5.76$ \end_inset for H3K4me2, and \begin_inset Formula $-4.00$ \end_inset for H3K27me3. \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout I also have some figures looking at interactions between marks (e.g. what if a promoter has both H3K4me3 and H3K27me3), but I don't know if that much detail is warranted here, since all the effects just seem approximate ly additive anyway. \end_layout \end_inset \end_layout \begin_layout Subsection Gene expression and promoter histone methylation patterns in naive and memory show convergence at day 14 \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash afterpage{ \end_layout \begin_layout Plain Layout \backslash begin{landscape} \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float table wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Tabular \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Number of significant promoters \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Est. differentially modified promoters \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Time Point \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout H3K4me2 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout H3K4me3 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout H3K27me3 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout H3K4me2 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout H3K4me3 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout H3K27me3 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Day 0 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 4553 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 927 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 6 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 9967 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 4149 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 2404 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Day 1 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 567 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 278 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 1570 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 4370 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 2145 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 6598 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Day 5 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 2313 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 139 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 490 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 9450 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 1148 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 4141 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Day 14 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0 \end_layout \end_inset \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "tab:Number-signif-promoters" \end_inset Number of differentially modified promoters between naive and memory cells at each time point after activation. \series default This table shows both the number of differentially modified promoters detected at a 10% FDR threshold (left half), and the total number of differentially modified promoters as estimated using the method of \begin_inset CommandInset citation LatexCommand cite key "Phipson2013" literal "false" \end_inset (right half). \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash end{landscape} \end_layout \begin_layout Plain Layout } \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float figure placement p wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K4me2-promoter-PCA-group-CROP.png lyxscale 25 width 45col% groupId pcoa-prom-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:PCoA-H3K4me2-prom" \end_inset PCoA plot of H3K4me2 promoters, after subtracting surrogate variables \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K4me3-promoter-PCA-group-CROP.png lyxscale 25 width 45col% groupId pcoa-prom-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:PCoA-H3K4me3-prom" \end_inset PCoA plot of H3K4me3 promoters, after subtracting surrogate variables \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K27me3-promoter-PCA-group-CROP.png lyxscale 25 width 45col% groupId pcoa-prom-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:PCoA-H3K27me3-prom" \end_inset PCoA plot of H3K27me3 promoters, after subtracting surrogate variables \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/RNA-seq/PCA-final-23-CROP.png lyxscale 25 width 45col% groupId pcoa-prom-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:RNA-PCA-group" \end_inset RNA-seq PCoA showing principal coordinates 2 and 3. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:PCoA-promoters" \end_inset PCoA plots for promoter ChIP-seq and expression RNA-seq data \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Check up on figure refs in this paragraph \end_layout \end_inset \end_layout \begin_layout Standard We hypothesized that if naive cells had differentiated into memory cells by Day 14, then their patterns of expression and histone modification should converge with those of memory cells at Day 14. Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:PCoA-promoters" plural "false" caps "false" noprefix "false" \end_inset shows the patterns of variation in all 3 histone marks in the promoter regions of the genome using principal coordinate analysis. All 3 marks show a noticeable convergence between the naive and memory samples at day 14, visible as an overlapping of the day 14 groups on each plot. This is consistent with the counts of significantly differentially modified promoters and estimates of the total numbers of differentially modified promoters shown in Table \begin_inset CommandInset ref LatexCommand ref reference "tab:Number-signif-promoters" plural "false" caps "false" noprefix "false" \end_inset . For all histone marks, evidence of differential modification between naive and memory samples was detected at every time point except day 14. The day 14 convergence pattern is also present in the RNA-seq data (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:RNA-PCA-group" plural "false" caps "false" noprefix "false" \end_inset ), albeit in the 2nd and 3rd principal coordinates, indicating that it is not the most dominant pattern driving gene expression. Taken together, the data show that promoter histone methylation for these 3 histone marks and RNA expression for naive and memory cells are most similar at day 14, the furthest time point after activation. MOFA was also able to capture this day 14 convergence pattern in latent factor 5 (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:mofa-lf-scatter" plural "false" caps "false" noprefix "false" \end_inset ), which accounts for shared variation across all 3 histone marks and the RNA-seq data, confirming that this convergence is a coordinated pattern across all 4 data sets. While this observation does not prove that the naive cells have differentiated into memory cells at Day 14, it is consistent with that hypothesis. \end_layout \begin_layout Subsection Effect of H3K4me2 and H3K4me3 promoter coverage upstream vs downstream of TSS \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Need a better section title, for this and the next one. \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Make sure use of coverage/abundance/whatever is consistent. \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout For the figures in this section and the next, the group labels are arbitrary, so if time allows, it would be good to manually reorder them in a logical way, e.g. most upstream to most downstream. If this is done, make sure to update the text with the correct group labels. \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash afterpage{ \end_layout \begin_layout Plain Layout \backslash begin{landscape} \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K4me2-neighborhood-clusters-CROP.png lyxscale 25 width 30col% groupId covprof-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:H3K4me2-neighborhood-clusters" \end_inset Average relative coverage for each bin in each cluster \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K4me2-neighborhood-PCA-CROP.png lyxscale 25 width 30col% groupId covprof-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:H3K4me2-neighborhood-pca" \end_inset PCA of relative coverage depth, colored by K-means cluster membership. \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K4me2-neighborhood-expression-CROP.png lyxscale 25 width 30col% groupId covprof-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:H3K4me2-neighborhood-expression" \end_inset Gene expression grouped by promoter coverage clusters. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:H3K4me2-neighborhood" \end_inset K-means clustering of promoter H3K4me2 relative coverage depth in naive day 0 samples. \series default H3K4me2 ChIP-seq reads were binned into 500-bp windows tiled across each promoter from 5 \begin_inset space ~ \end_inset kbp upstream to 5 \begin_inset space ~ \end_inset kbp downstream, and the logCPM values were normalized within each promoter to an average of 0, yielding relative coverage depths. These were then grouped using K-means clustering with \begin_inset Formula $K=6$ \end_inset , \series bold \series default and the average bin values were plotted for each cluster (a). The \begin_inset Formula $x$ \end_inset -axis is the genomic coordinate of each bin relative to the the transcription start site, and the \begin_inset Formula $y$ \end_inset -axis is the mean relative coverage depth of that bin across all promoters in the cluster. Each line represents the average \begin_inset Quotes eld \end_inset shape \begin_inset Quotes erd \end_inset of the promoter coverage for promoters in that cluster. PCA was performed on the same data, and the first two principal components were plotted, coloring each point by its K-means cluster identity (b). For each cluster, the distribution of gene expression values was plotted (c). \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash end{landscape} \end_layout \begin_layout Plain Layout } \end_layout \end_inset \end_layout \begin_layout Standard To test whether the position of a histone mark relative to a gene's transcriptio n start site (TSS) was important, we looked at the \begin_inset Quotes eld \end_inset landscape \begin_inset Quotes erd \end_inset of ChIP-seq read coverage in naive Day 0 samples within 5 kb of each gene's TSS by binning reads into 500-bp windows tiled across each promoter LogCPM values were calculated for the bins in each promoter and then the average logCPM for each promoter's bins was normalized to zero, such that the values represent coverage relative to other regions of the same promoter rather than being proportional to absolute read count. The promoters were then clustered based on the normalized bin abundances using \begin_inset Formula $k$ \end_inset -means clustering with \begin_inset Formula $K=6$ \end_inset . Different values of \begin_inset Formula $K$ \end_inset were also tested, but did not substantially change the interpretation of the data. \end_layout \begin_layout Standard For H3K4me2, plotting the average bin abundances for each cluster reveals a simple pattern (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me2-neighborhood-clusters" plural "false" caps "false" noprefix "false" \end_inset ): Cluster 5 represents a completely flat promoter coverage profile, likely consisting of genes with no H3K4me2 methylation in the promoter. All the other clusters represent a continuum of peak positions relative to the TSS. In order from must upstream to most downstream, they are Clusters 6, 4, 3, 1, and 2. There do not appear to be any clusters representing coverage patterns other than lone peaks, such as coverage troughs or double peaks. Next, all promoters were plotted in a PCA plot based on the same relative bin abundance data, and colored based on cluster membership (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me2-neighborhood-pca" plural "false" caps "false" noprefix "false" \end_inset ). The PCA plot shows Cluster 5 (the \begin_inset Quotes eld \end_inset no peak \begin_inset Quotes erd \end_inset cluster) at the center, with the other clusters arranged in a counter-clockwise arc around it in the order noted above, from most upstream peak to most downstream. Notably, the \begin_inset Quotes eld \end_inset clusters \begin_inset Quotes erd \end_inset form a single large \begin_inset Quotes eld \end_inset cloud \begin_inset Quotes erd \end_inset with no apparent separation between them, further supporting the conclusion that these clusters represent an arbitrary partitioning of a continuous distribution of promoter coverage landscapes. While the clusters are a useful abstraction that aids in visualization, they are ultimately not an accurate representation of the data. A better representation might be something like a polar coordinate system with the origin at the center of Cluster 5, where the radius represents the peak height above the background and the angle represents the peak's position upstream or downstream of the TSS. The continuous nature of the distribution also explains why different values of \begin_inset Formula $K$ \end_inset led to similar conclusions. \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout RNA-seq values in the plots use logCPM but should really use logFPKM or logTPM. Fix if time allows. \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Should have a table of p-values on difference of means between Cluster 5 and the others. \end_layout \end_inset \end_layout \begin_layout Standard To investigate the association between relative peak position and gene expressio n, we plotted the Naive Day 0 expression for the genes in each cluster (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me2-neighborhood-expression" plural "false" caps "false" noprefix "false" \end_inset ). Most genes in Cluster 5, the \begin_inset Quotes eld \end_inset no peak \begin_inset Quotes erd \end_inset cluster, have low expression values. Taking this as the \begin_inset Quotes eld \end_inset baseline \begin_inset Quotes erd \end_inset distribution when no H3K4me2 methylation is present, we can compare the other clusters' distributions to determine which peak positions are associated with elevated expression. As might be expected, the 3 clusters representing peaks closest to the TSS, Clusters 1, 3, and 4, show the highest average expression distributions. Specifically, these clusters all have their highest ChIP-seq abundance within 1kb of the TSS, consistent with the previously determined promoter radius. In contrast, cluster 6, which represents peaks several kb upstream of the TSS, shows a slightly higher average expression than baseline, while Cluster 2, which represents peaks several kb downstream, doesn't appear to show any appreciable difference. Interestingly, the cluster with the highest average expression is Cluster 1, which represents peaks about 1 kb downstream of the TSS, rather than Cluster 3, which represents peaks centered directly at the TSS. This suggests that conceptualizing the promoter as a region centered on the TSS with a certain \begin_inset Quotes eld \end_inset radius \begin_inset Quotes erd \end_inset may be an oversimplification – a peak that is a specific distance from the TSS may have a different degree of influence depending on whether it is upstream or downstream of the TSS. \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash afterpage{ \end_layout \begin_layout Plain Layout \backslash begin{landscape} \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K4me3-neighborhood-clusters-CROP.png lyxscale 25 width 30col% groupId covprof-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:H3K4me3-neighborhood-clusters" \end_inset Average relative coverage for each bin in each cluster \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K4me3-neighborhood-PCA-CROP.png lyxscale 25 width 30col% groupId covprof-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:H3K4me3-neighborhood-pca" \end_inset PCA of relative coverage depth, colored by K-means cluster membership. \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K4me3-neighborhood-expression-CROP.png lyxscale 25 width 30col% groupId covprof-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:H3K4me3-neighborhood-expression" \end_inset Gene expression grouped by promoter coverage clusters. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:H3K4me3-neighborhood" \end_inset K-means clustering of promoter H3K4me3 relative coverage depth in naive day 0 samples. \series default H3K4me2 ChIP-seq reads were binned into 500-bp windows tiled across each promoter from 5 \begin_inset space ~ \end_inset kbp upstream to 5 \begin_inset space ~ \end_inset kbp downstream, and the logCPM values were normalized within each promoter to an average of 0, yielding relative coverage depths. These were then grouped using K-means clustering with \begin_inset Formula $K=6$ \end_inset , \series bold \series default and the average bin values were plotted for each cluster (a). The \begin_inset Formula $x$ \end_inset -axis is the genomic coordinate of each bin relative to the the transcription start site, and the \begin_inset Formula $y$ \end_inset -axis is the mean relative coverage depth of that bin across all promoters in the cluster. Each line represents the average \begin_inset Quotes eld \end_inset shape \begin_inset Quotes erd \end_inset of the promoter coverage for promoters in that cluster. PCA was performed on the same data, and the first two principal components were plotted, coloring each point by its K-means cluster identity (b). For each cluster, the distribution of gene expression values was plotted (c). \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash end{landscape} \end_layout \begin_layout Plain Layout } \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Is there more to say here? \end_layout \end_inset \end_layout \begin_layout Standard All observations described above for H3K4me2 ChIP-seq also appear to hold for H3K4me3 as well (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me3-neighborhood" plural "false" caps "false" noprefix "false" \end_inset ). This is expected, since there is a high correlation between the positions where both histone marks occur. \end_layout \begin_layout Subsection Promoter coverage H3K27me3 \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash afterpage{ \end_layout \begin_layout Plain Layout \backslash begin{landscape} \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K27me3-neighborhood-clusters-CROP.png lyxscale 25 width 30col% groupId covprof-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:H3K27me3-neighborhood-clusters" \end_inset Average relative coverage for each bin in each cluster \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K27me3-neighborhood-PCA-CROP.png lyxscale 25 width 30col% groupId covprof-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:H3K27me3-neighborhood-pca" \end_inset PCA of relative coverage depth, colored by K-means cluster membership. \series default Note that Cluster 6 is hidden behind all the other clusters. \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/ChIP-seq/H3K27me3-neighborhood-expression-CROP.png lyxscale 25 width 30col% groupId covprof-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:H3K27me3-neighborhood-expression" \end_inset Gene expression grouped by promoter coverage clusters. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Repeated figure legends are kind of an issue here. What to do? \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:H3K27me3-neighborhood" \end_inset K-means clustering of promoter H3K27me3 relative coverage depth in naive day 0 samples. \series default H3K27me3 ChIP-seq reads were binned into 500-bp windows tiled across each promoter from 5 \begin_inset space ~ \end_inset kbp upstream to 5 \begin_inset space ~ \end_inset kbp downstream, and the logCPM values were normalized within each promoter to an average of 0, yielding relative coverage depths. These were then grouped using \begin_inset Formula $k$ \end_inset -means clustering with \begin_inset Formula $K=6$ \end_inset , \series bold \series default and the average bin values were plotted for each cluster (a). The \begin_inset Formula $x$ \end_inset -axis is the genomic coordinate of each bin relative to the the transcription start site, and the \begin_inset Formula $y$ \end_inset -axis is the mean relative coverage depth of that bin across all promoters in the cluster. Each line represents the average \begin_inset Quotes eld \end_inset shape \begin_inset Quotes erd \end_inset of the promoter coverage for promoters in that cluster. PCA was performed on the same data, and the first two principal components were plotted, coloring each point by its K-means cluster identity (b). For each cluster, the distribution of gene expression values was plotted (c). \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash end{landscape} \end_layout \begin_layout Plain Layout } \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Should maybe re-explain what was done or refer back to the previous section. \end_layout \end_inset \end_layout \begin_layout Standard Unlike both H3K4 marks, whose main patterns of variation appear directly related to the size and position of a single peak within the promoter, the patterns of H3K27me3 methylation in promoters are more complex (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K27me3-neighborhood" plural "false" caps "false" noprefix "false" \end_inset ). Once again looking at the relative coverage in a 500-bp wide bins in a 5kb radius around each TSS, promoters were clustered based on the normalized relative coverage values in each bin using \begin_inset Formula $k$ \end_inset -means clustering with \begin_inset Formula $K=6$ \end_inset (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K27me3-neighborhood-clusters" plural "false" caps "false" noprefix "false" \end_inset ). This time, 3 \begin_inset Quotes eld \end_inset axes \begin_inset Quotes erd \end_inset of variation can be observed, each represented by 2 clusters with opposing patterns. The first axis is greater upstream coverage (Cluster 1) vs. greater downstream coverage (Cluster 3); the second axis is the coverage at the TSS itself: peak (Cluster 4) or trough (Cluster 2); lastly, the third axis represents a trough upstream of the TSS (Cluster 5) vs. downstream of the TSS (Cluster 6). Referring to these opposing pairs of clusters as axes of variation is justified , because they correspond precisely to the first 3 principal components in the PCA plot of the relative coverage values (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K27me3-neighborhood-pca" plural "false" caps "false" noprefix "false" \end_inset ). The PCA plot reveals that as in the case of H3K4me2, all the \begin_inset Quotes eld \end_inset clusters \begin_inset Quotes erd \end_inset are really just sections of a single connected cloud rather than discrete clusters. The cloud is approximately ellipsoid-shaped, with each PC being an axis of the ellipse, and each cluster consisting of a pyramidal section of the ellipsoid. \end_layout \begin_layout Standard In Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K27me3-neighborhood-expression" plural "false" caps "false" noprefix "false" \end_inset , we can see that Clusters 1 and 2 are the only clusters with higher gene expression than the others. For Cluster 2, this is expected, since this cluster represents genes with depletion of H3K27me3 near the promoter. Hence, elevated expression in cluster 2 is consistent with the conventional view of H3K27me3 as a deactivating mark. However, Cluster 1, the cluster with the most elevated gene expression, represents genes with elevated coverage upstream of the TSS, or equivalently, decreased coverage downstream, inside the gene body. The opposite pattern, in which H3K27me3 is more abundant within the gene body and less abundance in the upstream promoter region, does not show any elevation in gene expression. As with H3K4me2, this shows that the location of H3K27 trimethylation relative to the TSS is potentially an important factor beyond simple proximity. \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Show the figures where the negative result ended this line of inquiry. I need to debug some errors resulting from an R upgrade to do this. \end_layout \end_inset \end_layout \begin_layout Subsection Defined pattern analysis \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout This was where I defined interesting expression patterns and then looked at initial relative promoter coverage for each expression pattern. Negative result. I forgot about this until recently. Worth including? Remember to also write methods. \end_layout \end_inset \end_layout \begin_layout Subsection Promoter CpG islands? \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status collapsed \begin_layout Plain Layout I forgot until recently about the work I did on this. Worth including? Remember to also write methods. \end_layout \end_inset \end_layout \begin_layout Section Discussion \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Write better section headers \end_layout \end_inset \end_layout \begin_layout Subsection Effective promoter radius \end_layout \begin_layout Standard Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:near-promoter-peak-enrich" plural "false" caps "false" noprefix "false" \end_inset shows that H3K4me2, H3K4me3, and H3K27me3 are all enriched near promoters, relative to the rest of the genome, consistent with their conventionally understood role in regulating gene transcription. Interestingly, the radius within this enrichment occurs is not the same for each histone mark. H3K4me2 and H3K4me3 are enriched within a 1 \begin_inset space \thinspace{} \end_inset kb radius, while H3K27me3 is enriched within 2.5 \begin_inset space \thinspace{} \end_inset kb. Notably, the determined promoter radius was consistent across all experimental conditions, varying only between different histone marks. This suggests that the conventional \begin_inset Quotes eld \end_inset one size fits all \begin_inset Quotes erd \end_inset approach of defining a single promoter region for each gene (or each TSS) and using that same promoter region for analyzing all types of genomic data within an experiment may not be appropriate, and a better approach may be to use a separate promoter radius for each kind of data, with each radius being derived from the data itself. Furthermore, the apparent asymmetry of upstream and downstream promoter histone modification with respect to gene expression, seen in Figures \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me2-neighborhood" plural "false" caps "false" noprefix "false" \end_inset , \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me3-neighborhood" plural "false" caps "false" noprefix "false" \end_inset , and \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K27me3-neighborhood" plural "false" caps "false" noprefix "false" \end_inset , shows that even the concept of a promoter \begin_inset Quotes eld \end_inset radius \begin_inset Quotes erd \end_inset is likely an oversimplification. At a minimum, nearby enrichment of peaks should be evaluated separately for both upstream and downstream peaks, and an appropriate \begin_inset Quotes eld \end_inset radius \begin_inset Quotes erd \end_inset should be selected for each direction. \end_layout \begin_layout Standard Figures \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me2-neighborhood" plural "false" caps "false" noprefix "false" \end_inset and \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me3-neighborhood" plural "false" caps "false" noprefix "false" \end_inset show that the determined promoter radius of 1 \begin_inset space ~ \end_inset kb is approximately consistent with the distance from the TSS at which enrichmen t of H3K4 methylation correlates with increased expression, showing that this radius, which was determined by a simple analysis of measuring the distance from each TSS to the nearest peak, also has functional significance. For H3K27me3, the correlation between histone modification near the promoter and gene expression is more complex, involving non-peak variations such as troughs in coverage at the TSS and asymmetric coverage upstream and downstream, so it is difficult in this case to evaluate whether the 2.5 \begin_inset space ~ \end_inset kb radius determined from TSS-to-peak distances is functionally significant. However, the two patterns of coverage associated with elevated expression levels both have interesting features within this radius. \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout My instinct is to say \begin_inset Quotes eld \end_inset further study is needed \begin_inset Quotes erd \end_inset here, but that goes in Chapter 5, right? \end_layout \end_inset \end_layout \begin_layout Subsection Convergence \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Look up some more references for these histone marks being involved in memory differentiation. (Ask Sarah) \end_layout \end_inset \end_layout \begin_layout Standard We have observed that all 3 histone marks and the gene expression data all exhibit evidence of convergence in abundance between naive and memory cells by day 14 after activation (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:PCoA-promoters" plural "false" caps "false" noprefix "false" \end_inset , Table \begin_inset CommandInset ref LatexCommand ref reference "tab:Number-signif-promoters" plural "false" caps "false" noprefix "false" \end_inset ). The MOFA latent factor scatter plots (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:mofa-lf-scatter" plural "false" caps "false" noprefix "false" \end_inset ) show that this pattern of convergence is captured in latent factor 5. Like all the latent factors in this plot, this factor explains a substantial portion of the variance in all 4 data sets, indicating a coordinated pattern of variation shared across all histone marks and gene expression. This, of course, is consistent with the expectation that any naive CD4 T-cells remaining at day 14 should have differentiated into memory cells by that time, and should therefore have a genomic state similar to memory cells. This convergence is evidence that these histone marks all play an important role in the naive-to-memory differentiation process. A histone mark that was not involved in naive-to-memory differentiation would not be expected to converge in this way after activation. \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/LaMere2016_fig8.pdf lyxscale 50 width 60col% groupId colwidth \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:Lamere2016-Fig8" \end_inset Lamere 2016 Figure 8 \begin_inset CommandInset citation LatexCommand cite key "LaMere2016" literal "false" \end_inset , \begin_inset Quotes eld \end_inset Model for the role of H3K4 methylation during CD4 T-cell activation. \begin_inset Quotes erd \end_inset \series default Reproduced with permission. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard In H3K4me2, H3K4me3, and RNA-seq, this convergence appears to be in progress already by Day 5, shown by the smaller distance between naive and memory cells at day 5 along the \begin_inset Formula $y$ \end_inset -axes in Figures \begin_inset CommandInset ref LatexCommand ref reference "fig:PCoA-H3K4me2-prom" plural "false" caps "false" noprefix "false" \end_inset , \begin_inset CommandInset ref LatexCommand ref reference "fig:PCoA-H3K4me3-prom" plural "false" caps "false" noprefix "false" \end_inset , and \begin_inset CommandInset ref LatexCommand ref reference "fig:RNA-PCA-group" plural "false" caps "false" noprefix "false" \end_inset . This agrees with the model proposed by Sarah Lamere based on an prior analysis of the same data, shown in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:Lamere2016-Fig8" plural "false" caps "false" noprefix "false" \end_inset , which shows the pattern of H3K4 methylation and expression for naive cells and memory cells converging at day 5. This model was developed without the benefit of the PCoA plots in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:PCoA-promoters" plural "false" caps "false" noprefix "false" \end_inset , which have been corrected for confounding factors by ComBat and SVA. This shows that proper batch correction assists in extracting meaningful patterns in the data while eliminating systematic sources of irrelevant variation in the data, allowing simple automated procedures like PCoA to reveal interesting behaviors in the data that were previously only detectable by a detailed manual analysis. \end_layout \begin_layout Standard While the ideal comparison to demonstrate this convergence would be naive cells at day 14 to memory cells at day 0, this is not feasible in this experimental system, since neither naive nor memory cells are able to fully return to their pre-activation state, as shown by the lack of overlap between days 0 and 14 for either naive or memory cells in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:PCoA-promoters" plural "false" caps "false" noprefix "false" \end_inset . \end_layout \begin_layout Subsection Positional \end_layout \begin_layout Standard When looking at patterns in the relative coverage of each histone mark near the TSS of each gene, several interesting patterns were apparent. For H3K4me2 and H3K4me3, the pattern was straightforward: the consistent pattern across all promoters was a single peak a few kb wide, with the main axis of variation being the position of this peak relative to the TSS (Figures \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me2-neighborhood" plural "false" caps "false" noprefix "false" \end_inset & \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me3-neighborhood" plural "false" caps "false" noprefix "false" \end_inset ). There were no obvious \begin_inset Quotes eld \end_inset preferred \begin_inset Quotes erd \end_inset positions, but rather a continuous distribution of relative positions ranging all across the promoter region. The association with gene expression was also straightforward: peaks closer to the TSS were more strongly associated with elevated gene expression. Coverage downstream of the TSS appears to be more strongly associated with elevated expression than coverage the same distance upstream, indicating that the \begin_inset Quotes eld \end_inset effective promoter region \begin_inset Quotes erd \end_inset for H3K4me2 and H3K4me3 may be centered downstream of the TSS. \end_layout \begin_layout Standard The relative promoter coverage for H3K27me3 had a more complex pattern, with two specific patterns of promoter coverage associated with elevated expression: a sharp depletion of H3K27me3 around the TSS relative to the surrounding area, and a depletion of H3K27me3 downstream of the TSS relative to upstream (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K27me3-neighborhood" plural "false" caps "false" noprefix "false" \end_inset ). A previous study found that H3K27me3 depletion within the gene body was associated with elevated gene expression in 4 different cell types in mice \begin_inset CommandInset citation LatexCommand cite key "Young2011" literal "false" \end_inset . This is consistent with the second pattern described here. This study also reported that a spike in coverage at the TSS was associated with \emph on lower \emph default expression, which is indirectly consistent with the first pattern described here, in the sense that it associates lower H3K27me3 levels near the TSS with higher expression. \end_layout \begin_layout Subsection Workflow \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash afterpage{ \end_layout \begin_layout Plain Layout \backslash begin{landscape} \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/CD4-csaw/rulegraphs/rulegraph-all.pdf lyxscale 50 width 100col% height 95theight% \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:rulegraph" \end_inset \series bold Dependency graph of steps in reproducible workflow. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash end{landscape} \end_layout \begin_layout Plain Layout } \end_layout \end_inset \end_layout \begin_layout Standard The analyses described in this chapter were organized into a reproducible workflow using the Snakemake workflow management system. As shown in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:rulegraph" plural "false" caps "false" noprefix "false" \end_inset , the workflow includes many steps with complex dependencies between them. For example, the step that counts the number of ChIP-seq reads in 500 \begin_inset space ~ \end_inset bp windows in each promoter (the starting point for Figures \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me2-neighborhood" plural "false" caps "false" noprefix "false" \end_inset , \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me3-neighborhood" plural "false" caps "false" noprefix "false" \end_inset , and \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K27me3-neighborhood" plural "false" caps "false" noprefix "false" \end_inset ), named \begin_inset Formula $\texttt{chipseq\_count\_tss\_neighborhoods}$ \end_inset , depends on the RNA-seq abundance estimates in order to select the most-used TSS for each gene, the aligned ChIP-seq reads, the index for those reads, and the blacklist of regions to be excluded from ChIP-seq analysis. Each step declares its inputs and outputs, and Snakemake uses these to determine the dependencies between steps. Each step is marked as depending on all the steps whose outputs match its inputs, generating the workflow graph in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:rulegraph" plural "false" caps "false" noprefix "false" \end_inset , which Snakemake uses to determine order in which to execute each step so that each step is executed only after all of the steps it depends on have completed, thereby automating the entire workflow from start to finish. \end_layout \begin_layout Standard In addition to simply making it easier to organize the steps in the analysis, structuring the analysis as a workflow allowed for some analysis strategies that would not have been practical otherwise. For example, 5 different RNA-seq quantification methods were tested against two different reference transcriptome annotations for a total of 10 different quantifications of the same RNA-seq data. These were then compared against each other in the exploratory data analysis step, to determine that the results were not very sensitive to either the choice of quantification method or the choice of annotation. This was possible with a single script for the exploratory data analysis, because Snakemake was able to automate running this script for every combinatio n of method and reference. In a similar manner, two different peak calling methods were tested against each other, and in this case it was determined that SICER was unambiguously superior to MACS for all histone marks studied. By enabling these types of comparisons, structuring the analysis as an automated workflow allowed important analysis decisions to be made in a data-driven way, by running every reasonable option through the downstream steps, seeing the consequences of choosing each option, and deciding accordingl y. \end_layout \begin_layout Subsection Data quality issues limit conclusions \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Is this needed? \end_layout \end_inset \end_layout \begin_layout Section Future Directions \end_layout \begin_layout Standard The analysis of RNA-seq and ChIP-seq in CD4 T-cells in Chapter 2 is in many ways a preliminary study that suggests a multitude of new avenues of investigat ion. Here we consider a selection of such avenues. \end_layout \begin_layout Subsection Negative results \end_layout \begin_layout Standard Two additional analyses were conducted beyond those reported in the results. First, we searched for evidence that the presence or absence of a CpG island in the promoter was correlated with increases or decreases in gene expression or any histone mark in any of the tested contrasts. Second, we searched for evidence that the relative ChIP-seq coverage profiles prior to activations could predict the change in expression of a gene after activation. Neither analysis turned up any clear positive results. \end_layout \begin_layout Subsection Improve on the idea of an effective promoter radius \end_layout \begin_layout Standard This study introduced the concept of an \begin_inset Quotes eld \end_inset effective promoter radius \begin_inset Quotes erd \end_inset specific to each histone mark based on distance from the TSS within which an excess of peaks was called for that mark. This concept was then used to guide further analyses throughout the study. However, while the effective promoter radius was useful in those analyses, it is both limited in theory and shown in practice to be a possible oversimplif ication. First, the effective promoter radii used in this study were chosen based on manual inspection of the TSS-to-peak distance distributions in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:near-promoter-peak-enrich" plural "false" caps "false" noprefix "false" \end_inset , selecting round numbers of analyst convenience (Table \begin_inset CommandInset ref LatexCommand ref reference "tab:effective-promoter-radius" plural "false" caps "false" noprefix "false" \end_inset ). It would be better to define an algorithm that selects a more precise radius based on the features of the graph. One possible way to do this would be to randomly rearrange the called peaks throughout the genome many (while preserving the distribution of peak widths) and re-generate the same plot as in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:near-promoter-peak-enrich" plural "false" caps "false" noprefix "false" \end_inset . This would yield a better \begin_inset Quotes eld \end_inset background \begin_inset Quotes erd \end_inset distribution that demonstrates the degree of near-TSS enrichment that would be expected by random chance. The effective promoter radius could be defined as the point where the true distribution diverges from the randomized background distribution. \end_layout \begin_layout Standard Furthermore, the above definition of effective promoter radius has the significa nt limitation of being based on the peak calling method. It is thus very sensitive to the choice of peak caller and significance threshold for calling peaks, as well as the degree of saturation in the sequencing. Calling peaks from ChIP-seq samples with insufficient coverage depth, with the wrong peak caller, or with a different significance threshold could give a drastically different number of called peaks, and hence a drastically different distribution of peak-to-TSS distances. To address this, it is desirable to develop a better method of determining the effective promoter radius that relies only on the distribution of read coverage around the TSS, independent of the peak calling. Furthermore, as demonstrated by the upstream-downstream asymmetries observed in Figures \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me2-neighborhood" plural "false" caps "false" noprefix "false" \end_inset , \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me3-neighborhood" plural "false" caps "false" noprefix "false" \end_inset , and \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K27me3-neighborhood" plural "false" caps "false" noprefix "false" \end_inset , this definition should determine a different radius for the upstream and downstream directions. At this point, it may be better to rename this concept \begin_inset Quotes eld \end_inset effective promoter extent \begin_inset Quotes erd \end_inset and avoid the word \begin_inset Quotes eld \end_inset radius \begin_inset Quotes erd \end_inset , since a radius implies a symmetry about the TSS that is not supported by the data. \end_layout \begin_layout Standard Beyond improving the definition of effective promoter extent, functional validation is necessary to show that this measure of near-TSS enrichment has biological meaning. Figures \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me2-neighborhood" plural "false" caps "false" noprefix "false" \end_inset and \begin_inset CommandInset ref LatexCommand ref reference "fig:H3K4me3-neighborhood" plural "false" caps "false" noprefix "false" \end_inset already provide a very limited functional validation of the chosen promoter extents for H3K4me2 and H3K4me3 by showing that spikes in coverage within this region are most strongly correlated with elevated gene expression. However, there are other ways to show functional relevance of the promoter extent. For example, correlations could be computed between read counts in peaks nearby gene promoters and the expression level of those genes, and these correlations could be plotted against the distance of the peak upstream or downstream of the gene's TSS. If the promoter extent truly defines a \begin_inset Quotes eld \end_inset sphere of influence \begin_inset Quotes erd \end_inset within which a histone mark is involved with the regulation of a gene, then the correlations for peaks within this extent should be significantly higher than those further upstream or downstream. Peaks within these extents may also be more likely to show differential modification than those outside genic regions of the genome. \end_layout \begin_layout Subsection Design experiments to focus on post-activation convergence of naive & memory cells \end_layout \begin_layout Standard In this study, a convergence between naive and memory cells was observed in both the pattern of gene expression and in epigenetic state of the 3 histone marks studied, consistent with the hypothesis that any naive cells remaining 14 days after activation have differentiated into memory cells, and that both gene expression and these histone marks are involved in this differentiation. However, the current study was not designed with this specific hypothesis in mind, and it therefore has some deficiencies with regard to testing it. The memory CD4 samples at day 14 do not resemble the memory samples at day 0, indicating that in the specific model of activation used for this experiment, the cells are not guaranteed to return to their original pre-activa tion state, or perhaps this process takes substantially longer than 14 days. This is a challenge for the convergence hypothesis because the ideal comparison to prove that naive cells are converging to a resting memory state would be to compare the final naive time point to the Day 0 memory samples, but this comparison is only meaningful if memory cells generally return to the same \begin_inset Quotes eld \end_inset resting \begin_inset Quotes erd \end_inset state that they started at. \end_layout \begin_layout Standard To better study the convergence hypothesis, a new experiment should be designed using a model system for T-cell activation that is known to allow cells to return as closely as possible to their pre-activation state. Alternatively, if it is not possible to find or design such a model system, the same cell cultures could be activated serially multiple times, and sequenced after each activation cycle right before the next activation. It is likely that several activations in the same model system will settle into a cyclical pattern, converging to a consistent \begin_inset Quotes eld \end_inset resting \begin_inset Quotes erd \end_inset state after each activation, even if this state is different from the initial resting state at Day 0. If so, it will be possible to compare the final states of both naive and memory cells to show that they converge despite different initial conditions. \end_layout \begin_layout Standard In addition, if naive-to-memory convergence is a general pattern, it should also be detectable in other epigenetic marks, including other histone marks and DNA methylation. An experiment should be designed studying a large number of epigenetic marks known or suspected to be involved in regulation of gene expression, assaying all of these at the same pre- and post-activation time points. Multi-dataset factor analysis methods like MOFA can then be used to identify coordinated patterns of regulation shared across many epigenetic marks. If possible, some \begin_inset Quotes eld \end_inset negative control \begin_inset Quotes erd \end_inset marks should be included that are known \emph on not \emph default to be involved in T-cell activation or memory formation. Of course, CD4 T-cells are not the only adaptive immune cells with memory. A similar study could be designed for CD8 T-cells, B-cells, and even specific subsets of CD4 T-cells. \end_layout \begin_layout Subsection Follow up on hints of interesting patterns in promoter relative coverage profiles \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout I think I might need to write up the negative results for the Promoter CpG and defined pattern analysis before writing this section. \end_layout \end_inset \end_layout \begin_layout Itemize Also find better normalizations: maybe borrow from MACS/SICER background correction methods? \end_layout \begin_layout Itemize For H3K4, define polar coordinates based on PC1 & 2: R = peak size, Theta = peak position. Then correlate with expression. \end_layout \begin_layout Itemize Current analysis only at Day 0. Need to study across time points. \end_layout \begin_layout Itemize Integrating data across so many dimensions is a significant analysis challenge \end_layout \begin_layout Subsection Investigate causes of high correlation between mutually exclusive histone marks \end_layout \begin_layout Standard The high correlation between coverage depth observed between H3K4me2 and H3K4me3 is both expected and unexpected. Since both marks are associated with elevated gene transcription, a positive correlation between them is not surprising. However, these two marks represent different post-translational modifications of the \emph on same \emph default lysine residue on the histone H3 polypeptide, which means that they cannot both be present on the same H3 subunit. Thus, the high correlation between them has several potential explanations. One possible reason is cell population heterogeneity: perhaps some genomic loci are frequently marked with H3K4me2 in some cells, while in other cells the same loci are marked with H3K4me3. Another possibility is allele-specific modifications: the loci are marked in each diploid cell with H3K4me2 on one allele and H3K4me3 on the other allele. Lastly, since each histone octamer contains 2 H3 subunits, it is possible that having one H3K4me2 mark and one H3K4me3 mark on a given histone octamer represents a distinct epigenetic state with a different function than either double H3K4me2 or double H3K4me3. \end_layout \begin_layout Standard These three hypotheses could be disentangled by single-cell ChIP-seq. If the correlation between these two histone marks persists even within the reads for each individual cell, then cell population heterogeneity cannot explain the correlation. Allele-specific modification can be tested for by looking at the correlation between read coverage of the two histone marks at heterozygous loci. If the correlation between read counts for opposite loci is low, then this is consistent with allele-specific modification. Finally if the modifications do not separate by either cell or allele, the colocation of these two marks is most likely occurring at the level of individual histones, with the heterogeneously modified histone representing a distinct state. \end_layout \begin_layout Standard However, another experiment would be required to show direct evidence of such a heterogeneously modified state. Specifically a \begin_inset Quotes eld \end_inset double ChIP \begin_inset Quotes erd \end_inset experiment would need to be performed, where the input DNA is first subjected to an immunoprecipitation pulldown from the anti-H3K4me2 antibody, and then the enriched material is collected, with proteins still bound, and immunoprecipitated \emph on again \emph default using the anti-H3K4me3 antibody. If this yields significant numbers of non-artifactual reads in the same regions as the individual pulldowns of the two marks, this is strong evidence that the two marks are occurring on opposite H3 subunits of the same histones. \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Try to see if double ChIP-seq is actually feasible, and if not, come up with some other idea for directly detecting the mixed mod state. Oh! Actually ChIP-seq isn't required, only double ChIP followed by quantificati on. That's one possible angle. \end_layout \end_inset \end_layout \begin_layout Chapter Improving array-based diagnostics for transplant rejection by optimizing data preprocessing \end_layout \begin_layout Standard \begin_inset Note Note status open \begin_layout Plain Layout Chapter author list: Me, Sunil, Tom, Padma, Dan \end_layout \end_inset \end_layout \begin_layout Section Approach \end_layout \begin_layout Subsection Proper pre-processing is essential for array data \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout This section could probably use some citations \end_layout \end_inset \end_layout \begin_layout Standard Microarrays, bead arrays, and similar assays produce raw data in the form of fluorescence intensity measurements, with the each intensity measurement proportional to the abundance of some fluorescently labelled target DNA or RNA sequence that base pairs to a specific probe sequence. However, these measurements for each probe are also affected my many technical confounding factors, such as the concentration of target material, strength of off-target binding, and the sensitivity of the imaging sensor. Some array designs also use multiple probe sequences for each target. Hence, extensive pre-processing of array data is necessary to normalize out the effects of these technical factors and summarize the information from multiple probes to arrive at a single usable estimate of abundance or other relevant quantity, such as a ratio of two abundances, for each target. \end_layout \begin_layout Standard The choice of pre-processing algorithms used in the analysis of an array data set can have a large effect on the results of that analysis. However, despite their importance, these steps are often neglected or rushed in order to get to the more scientifically interesting analysis steps involving the actual biology of the system under study. Hence, it is often possible to achieve substantial gains in statistical power, model goodness-of-fit, or other relevant performance measures, by checking the assumptions made by each preprocessing step and choosing specific normalization methods tailored to the specific goals of the current analysis. \end_layout \begin_layout Subsection Clinical diagnostic applications for microarrays require single-channel normalization \end_layout \begin_layout Standard As the cost of performing microarray assays falls, there is increasing interest in using genomic assays for diagnostic purposes, such as distinguishing healthy transplants (TX) from transplants undergoing acute rejection (AR) or acute dysfunction with no rejection (ADNR). However, the the standard normalization algorithm used for microarray data, Robust Multi-chip Average (RMA) \begin_inset CommandInset citation LatexCommand cite key "Irizarry2003a" literal "false" \end_inset , is not applicable in a clinical setting. Two of the steps in RMA, quantile normalization and probe summarization by median polish, depend on every array in the data set being normalized. This means that adding or removing any arrays from a data set changes the normalized values for all arrays, and data sets that have been normalized separately cannot be compared to each other. Hence, when using RMA, any arrays to be analyzed together must also be normalized together, and the set of arrays included in the data set must be held constant throughout an analysis. \end_layout \begin_layout Standard These limitations present serious impediments to the use of arrays as a diagnostic tool. When training a classifier, the samples to be classified must not be involved in any step of the training process, lest their inclusion bias the training process. Once a classifier is deployed in a clinical setting, the samples to be classified will not even \emph on exist \emph default at the time of training, so including them would be impossible even if it were statistically justifiable. Therefore, any machine learning application for microarrays demands that the normalized expression values computed for an array must depend only on information contained within that array. This would ensure that each array's normalization is independent of every other array, and that arrays normalized separately can still be compared to each other without bias. Such a normalization is commonly referred to as \begin_inset Quotes eld \end_inset single-channel normalization \begin_inset Quotes erd \end_inset . \end_layout \begin_layout Standard Frozen RMA (fRMA) addresses these concerns by replacing the quantile normalizati on and median polish with alternatives that do not introduce inter-array dependence, allowing each array to be normalized independently of all others \begin_inset CommandInset citation LatexCommand cite key "McCall2010" literal "false" \end_inset . Quantile normalization is performed against a pre-generated set of quantiles learned from a collection of 850 publicly available arrays sampled from a wide variety of tissues in the Gene Expression Omnibus (GEO). Each array's probe intensity distribution is normalized against these pre-gener ated quantiles. The median polish step is replaced with a robust weighted average of probe intensities, using inverse variance weights learned from the same public GEO data. The result is a normalization that satisfies the requirements mentioned above: each array is normalized independently of all others, and any two normalized arrays can be compared directly to each other. \end_layout \begin_layout Standard One important limitation of fRMA is that it requires a separate reference data set from which to learn the parameters (reference quantiles and probe weights) that will be used to normalize each array. These parameters are specific to a given array platform, and pre-generated parameters are only provided for the most common platforms, such as Affymetrix hgu133plus2. For a less common platform, such as hthgu133pluspm, is is necessary to learn custom parameters from in-house data before fRMA can be used to normalize samples on that platform \begin_inset CommandInset citation LatexCommand cite key "McCall2011" literal "false" \end_inset . \end_layout \begin_layout Standard One other option is the aptly-named Single Channel Array Normalization (SCAN), which adapts a normalization method originally designed for tiling arrays \begin_inset CommandInset citation LatexCommand cite key "Piccolo2012" literal "false" \end_inset . SCAN is truly single-channel in that it does not require a set of normalization parameters estimated from an external set of reference samples like fRMA does. \end_layout \begin_layout Subsection Heteroskedasticity must be accounted for in methylation array data \end_layout \begin_layout Standard DNA methylation arrays are a relatively new kind of assay that uses microarrays to measure the degree of methylation on cytosines in specific regions arrayed across the genome. First, bisulfite treatment converts all unmethylated cytosines to uracil (which then become thymine after amplification) while leaving methylated cytosines unaffected. Then, each target region is interrogated with two probes: one binds to the original genomic sequence and interrogates the level of methylated DNA, and the other binds to the same sequence with all cytosines replaced by thymidines and interrogates the level of unmethylated DNA. \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/methylvoom/sigmoid.pdf lyxscale 50 width 60col% groupId colwidth \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:Sigmoid-beta-m-mapping" \end_inset \series bold Sigmoid shape of the mapping between β and M values \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard After normalization, these two probe intensities are summarized in one of two ways, each with advantages and disadvantages. β \series bold \series default values, interpreted as fraction of DNA copies methylated, range from 0 to 1. β \series bold \series default values are conceptually easy to interpret, but the constrained range makes them unsuitable for linear modeling, and their error distributions are highly non-normal, which also frustrates linear modeling. M-values, interpreted as the log ratio of methylated to unmethylated copies, are computed by mapping the beta values from \begin_inset Formula $[0,1]$ \end_inset onto \begin_inset Formula $(-\infty,+\infty)$ \end_inset using a sigmoid curve (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:Sigmoid-beta-m-mapping" plural "false" caps "false" noprefix "false" \end_inset ). This transformation results in values with better statistical properties: the unconstrained range is suitable for linear modeling, and the error distributions are more normal. Hence, most linear modeling and other statistical testing on methylation arrays is performed using M-values. \end_layout \begin_layout Standard However, the steep slope of the sigmoid transformation near 0 and 1 tends to over-exaggerate small differences in β values near those extremes, which in turn amplifies the error in those values, leading to a U-shaped trend in the mean-variance curve: extreme values have higher variances than values near the middle. This mean-variance dependency must be accounted for when fitting the linear model for differential methylation, or else the variance will be systematically overestimated for probes with moderate M-values and underestimated for probes with extreme M-values. This is particularly undesirable for methylation data because the intermediate M-values are the ones of most interest, since they are more likely to represent areas of varying methylation, whereas extreme M-values typically represent complete methylation or complete lack of methylation. \end_layout \begin_layout Standard RNA-seq read count data are also known to show heteroskedasticity, and the voom method was introduced for modeling this heteroskedasticity by estimating the mean-variance trend in the data and using this trend to assign precision weights to each observation \begin_inset CommandInset citation LatexCommand cite key "Law2013" literal "false" \end_inset . While methylation array data are not derived from counts and have a very different mean-variance relationship from that of typical RNA-seq data, the voom method makes no specific assumptions on the shape of the mean-variance relationship – it only assumes that the relationship can be modeled as a smooth curve. Hence, the method is sufficiently general to model the mean-variance relationsh ip in methylation array data. However, the standard implementation of voom assumes that the input is given in raw read counts, and it must be adapted to run on methylation M-values. \end_layout \begin_layout Section Methods \end_layout \begin_layout Subsection Evaluation of classifier performance with different normalization methods \end_layout \begin_layout Standard For testing different expression microarray normalizations, a data set of 157 hgu133plus2 arrays was used, consisting of blood samples from kidney transplant patients whose grafts had been graded as TX, AR, or ADNR via biopsy and histology (46 TX, 69 AR, 42 ADNR) \begin_inset CommandInset citation LatexCommand cite key "Kurian2014" literal "true" \end_inset . Additionally, an external validation set of 75 samples was gathered from public GEO data (37 TX, 38 AR, no ADNR). \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Find appropriate GEO identifiers if possible. Kurian 2014 says GSE15296, but this seems to be different data. I also need to look up the GEO accession for the external validation set. \end_layout \end_inset \end_layout \begin_layout Standard To evaluate the effect of each normalization on classifier performance, the same classifier training and validation procedure was used after each normalization method. The PAM package was used to train a nearest shrunken centroid classifier on the training set and select the appropriate threshold for centroid shrinking. Then the trained classifier was used to predict the class probabilities of each validation sample. From these class probabilities, ROC curves and area-under-curve (AUC) values were generated \begin_inset CommandInset citation LatexCommand cite key "Turck2011" literal "false" \end_inset . Each normalization was tested on two different sets of training and validation samples. For internal validation, the 115 TX and AR arrays in the internal set were split at random into two equal sized sets, one for training and one for validation, each containing the same numbers of TX and AR samples as the other set. For external validation, the full set of 115 TX and AR samples were used as a training set, and the 75 external TX and AR samples were used as the validation set. Thus, 2 ROC curves and AUC values were generated for each normalization method: one internal and one external. Because the external validation set contains no ADNR samples, only classificati on of TX and AR samples was considered. The ADNR samples were included during normalization but excluded from all classifier training and validation. This ensures that the performance on internal and external validation sets is directly comparable, since both are performing the same task: distinguishing TX from AR. \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Summarize the get.best.threshold algorithm for PAM threshold selection, or just put the code online? \end_layout \end_inset \end_layout \begin_layout Standard Six different normalization strategies were evaluated. First, 2 well-known non-single-channel normalization methods were considered: RMA and dChip \begin_inset CommandInset citation LatexCommand cite key "Li2001,Irizarry2003a" literal "false" \end_inset . Since RMA produces expression values on a log2 scale and dChip does not, the values from dChip were log2 transformed after normalization. Next, RMA and dChip followed by Global Rank-invariant Set Normalization (GRSN) were tested \begin_inset CommandInset citation LatexCommand cite key "Pelz2008" literal "false" \end_inset . Post-processing with GRSN does not turn RMA or dChip into single-channel methods, but it may help mitigate batch effects and is therefore useful as a benchmark. Lastly, the two single-channel normalization methods, fRMA and SCAN, were tested \begin_inset CommandInset citation LatexCommand cite key "McCall2010,Piccolo2012" literal "false" \end_inset . When evaluating internal validation performance, only the 157 internal samples were normalized; when evaluating external validation performance, all 157 internal samples and 75 external samples were normalized together. \end_layout \begin_layout Standard For demonstrating the problem with separate normalization of training and validation data, one additional normalization was performed: the internal and external sets were each normalized separately using RMA, and the normalized data for each set were combined into a single set with no further attempts at normalizing between the two sets. The represents approximately how RMA would have to be used in a clinical setting, where the samples to be classified are not available at the time the classifier is trained. \end_layout \begin_layout Subsection Generating custom fRMA vectors for hthgu133pluspm array platform \end_layout \begin_layout Standard In order to enable fRMA normalization for the hthgu133pluspm array platform, custom fRMA normalization vectors were trained using the \begin_inset Flex Code status open \begin_layout Plain Layout frmaTools \end_layout \end_inset package \begin_inset CommandInset citation LatexCommand cite key "McCall2011" literal "false" \end_inset . Separate vectors were created for two types of samples: kidney graft biopsy samples and blood samples from graft recipients. For training, a 341 kidney biopsy samples from 2 data sets and 965 blood samples from 5 data sets were used as the reference set. Arrays were groups into batches based on unique combinations of sample type (blood or biopsy), diagnosis (TX, AR, etc.), data set, and scan date. Thus, each batch represents arrays of the same kind that were run together on the same day. For estimating the probe inverse variance weights, frmaTools requires equal-siz ed batches, which means a batch size must be chosen, and then batches smaller than that size must be ignored, while batches larger than the chosen size must be downsampled. This downsampling is performed randomly, so the sampling process is repeated 5 times and the resulting normalizations are compared to each other. \end_layout \begin_layout Standard To evaluate the consistency of the generated normalization vectors, the 5 fRMA vector sets generated from 5 random batch samplings were each used to normalize the same 20 randomly selected samples from each tissue. Then the normalized expression values for each probe on each array were compared across all normalizations. Each fRMA normalization was also compared against the normalized expression values obtained by normalizing the same 20 samples with ordinary RMA. \end_layout \begin_layout Subsection Modeling methylation array M-value heteroskedasticy in linear models with modified voom implementation \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Put code on Github and reference it. \end_layout \end_inset \end_layout \begin_layout Standard To investigate the whether DNA methylation could be used to distinguish between healthy and dysfunctional transplants, a data set of 78 Illumina 450k methylation arrays from human kidney graft biopsies was analyzed for differential methylation between 4 transplant statuses: healthy transplant (TX), transplants undergoing acute rejection (AR), acute dysfunction with no rejection (ADNR), and chronic allograft nephropathy (CAN). The data consisted of 33 TX, 9 AR, 8 ADNR, and 28 CAN samples. The uneven group sizes are a result of taking the biopsy samples before the eventual fate of the transplant was known. Each sample was additionally annotated with a donor ID (anonymized), Sex, Age, Ethnicity, Creatinine Level, and Diabetes diagnosis (all samples in this data set came from patients with either Type 1 or Type 2 diabetes). \end_layout \begin_layout Standard The intensity data were first normalized using subset-quantile within array normalization (SWAN) \begin_inset CommandInset citation LatexCommand cite key "Maksimovic2012" literal "false" \end_inset , then converted to intensity ratios (beta values) \begin_inset CommandInset citation LatexCommand cite key "Aryee2014" literal "false" \end_inset . Any probes binding to loci that overlapped annotated SNPs were dropped, and the annotated sex of each sample was verified against the sex inferred from the ratio of median probe intensities for the X and Y chromosomes. Then, the ratios were transformed to M-values. \end_layout \begin_layout Standard \begin_inset Float table wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Tabular \begin_inset Text \begin_layout Plain Layout Analysis \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout random effect \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout eBayes \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout SVA \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout weights \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout voom \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout A \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Yes \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Yes \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout No \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout No \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout No \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout B \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Yes \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Yes \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Yes \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Yes \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout No \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout C \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Yes \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Yes \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Yes \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Yes \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Yes \end_layout \end_inset \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "tab:Summary-of-meth-analysis" \end_inset Summary of analysis variants for methylation array data. \series default Each analysis included a different set of steps to adjust or account for various systematic features of the data. Random effect: The model included a random effect accounting for correlation between samples from the same patient \begin_inset CommandInset citation LatexCommand cite key "Smyth2005a" literal "false" \end_inset ; eBayes: Empirical bayes squeezing of per-probe variances toward the mean-varia nce trend \begin_inset CommandInset citation LatexCommand cite key "Ritchie2015" literal "false" \end_inset ; SVA: Surrogate variable analysis to account for unobserved confounders \begin_inset CommandInset citation LatexCommand cite key "Leek2007" literal "false" \end_inset ; Weights: Estimate sample weights to account for differences in sample quality \begin_inset CommandInset citation LatexCommand cite key "Liu2015,Ritchie2006" literal "false" \end_inset ; voom: Use mean-variance trend to assign individual sample weights \begin_inset CommandInset citation LatexCommand cite key "Law2013" literal "false" \end_inset . See the text for a more detailed explanation of each step. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard From the M-values, a series of parallel analyses was performed, each adding additional steps into the model fit to accommodate a feature of the data (see Table \begin_inset CommandInset ref LatexCommand ref reference "tab:Summary-of-meth-analysis" plural "false" caps "false" noprefix "false" \end_inset ). For analysis A, a \begin_inset Quotes eld \end_inset basic \begin_inset Quotes erd \end_inset linear modeling analysis was performed, compensating for known confounders by including terms for the factor of interest (transplant status) as well as the known biological confounders: sex, age, ethnicity, and diabetes. Since some samples came from the same patients at different times, the intra-patient correlation was modeled as a random effect, estimating a shared correlation value across all probes \begin_inset CommandInset citation LatexCommand cite key "Smyth2005a" literal "false" \end_inset . Then the linear model was fit, and the variance was modeled using empirical Bayes squeezing toward the mean-variance trend \begin_inset CommandInset citation LatexCommand cite key "Ritchie2015" literal "false" \end_inset . Finally, t-tests or F-tests were performed as appropriate for each test: t-tests for single contrasts, and F-tests for multiple contrasts. P-values were corrected for multiple testing using the Benjamini-Hochberg procedure for FDR control \begin_inset CommandInset citation LatexCommand cite key "Benjamini1995" literal "false" \end_inset . \end_layout \begin_layout Standard For the analysis B, surrogate variable analysis (SVA) was used to infer additional unobserved sources of heterogeneity in the data \begin_inset CommandInset citation LatexCommand cite key "Leek2007" literal "false" \end_inset . These surrogate variables were added to the design matrix before fitting the linear model. In addition, sample quality weights were estimated from the data and used during linear modeling to down-weight the contribution of highly variable arrays while increasing the weight to arrays with lower variability \begin_inset CommandInset citation LatexCommand cite key "Ritchie2006" literal "false" \end_inset . The remainder of the analysis proceeded as in analysis A. For analysis C, the voom method was adapted to run on methylation array data and used to model and correct for the mean-variance trend using individual observation weights \begin_inset CommandInset citation LatexCommand cite key "Law2013" literal "false" \end_inset , which were combined with the sample weights \begin_inset CommandInset citation LatexCommand cite key "Liu2015,Ritchie2006" literal "false" \end_inset . Each time weights were used, they were estimated once before estimating the random effect correlation value, and then the weights were re-estimated taking the random effect into account. The remainder of the analysis proceeded as in analysis B. \end_layout \begin_layout Section Results \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Improve subsection titles in this section. \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Reconsider subsection organization? \end_layout \end_inset \end_layout \begin_layout Subsection Separate normalization with RMA introduces unwanted biases in classification \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/PAM/predplot.pdf lyxscale 50 width 60col% groupId colwidth \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:Classifier-probabilities-RMA" \end_inset \series bold Classifier probabilities on validation samples when normalized with RMA together vs. separately. \series default The PAM classifier algorithm was trained on the training set of arrays to distinguish AR from TX and then used to assign class probabilities to the validation set. The process was performed after normalizing all samples together and after normalizing the training and test sets separately, and the class probabilities assigned to each sample in the validation set were plotted against each other (PP(AR), posterior probability of being AR). The color of each point indicates the true classification of that sample. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard To demonstrate the problem with non-single-channel normalization methods, we considered the problem of training a classifier to distinguish TX from AR using the samples from the internal set as training data, evaluating performance on the external set. First, training and evaluation were performed after normalizing all array samples together as a single set using RMA, and second, the internal samples were normalized separately from the external samples and the training and evaluation were repeated. For each sample in the validation set, the classifier probabilities from both classifiers were plotted against each other (Fig. \begin_inset CommandInset ref LatexCommand ref reference "fig:Classifier-probabilities-RMA" plural "false" caps "false" noprefix "false" \end_inset ). As expected, separate normalization biases the classifier probabilities, resulting in several misclassifications. In this case, the bias from separate normalization causes the classifier to assign a lower probability of AR to every sample. \end_layout \begin_layout Subsection fRMA and SCAN maintain classification performance while eliminating dependence on normalization strategy \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Float figure placement tb wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/PAM/ROC-TXvsAR-internal.pdf lyxscale 50 height 40theight% groupId roc-pam \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:ROC-PAM-int" \end_inset ROC curves for PAM on internal validation data \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \begin_inset Float figure placement tb wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/PAM/ROC-TXvsAR-external.pdf lyxscale 50 height 40theight% groupId roc-pam \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:ROC-PAM-ext" \end_inset ROC curves for PAM on external validation data \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:ROC-PAM-main" \end_inset ROC curves for PAM using different normalization strategies. \series default ROC curves were generated for PAM classification of AR vs TX after 6 different normalization strategies applied to the same data sets. Only fRMA and SCAN are single-channel normalizations. The other normalizations are for comparison. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float table wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Tabular \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none Normalization \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Single-channel? \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none Internal Val. AUC \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout External Val. AUC \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none RMA \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout No \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 0.852 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 0.713 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none dChip \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout No \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 0.891 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 0.657 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none RMA + GRSN \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout No \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 0.816 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 0.750 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none dChip + GRSN \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout No \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 0.875 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 0.642 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none fRMA \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Yes \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 0.863 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 0.718 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none SCAN \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Yes \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 0.853 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 0.689 \end_layout \end_inset \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "tab:AUC-PAM" \end_inset \series bold ROC curve AUC values for internal and external validation with 6 different normalization strategies. \series default These AUC values correspond to the ROC curves in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:ROC-PAM-main" plural "false" caps "false" noprefix "false" \end_inset . \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard For internal validation, the 6 methods' AUC values ranged from 0.816 to 0.891, as shown in Table \begin_inset CommandInset ref LatexCommand ref reference "tab:AUC-PAM" plural "false" caps "false" noprefix "false" \end_inset . Among the non-single-channel normalizations, dChip outperformed RMA, while GRSN reduced the AUC values for both dChip and RMA. Both single-channel methods, fRMA and SCAN, slightly outperformed RMA, with fRMA ahead of SCAN. However, the difference between RMA and fRMA is still quite small. Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:ROC-PAM-int" plural "false" caps "false" noprefix "false" \end_inset shows that the ROC curves for RMA, dChip, and fRMA look very similar and relatively smooth, while both GRSN curves and the curve for SCAN have a more jagged appearance. \end_layout \begin_layout Standard For external validation, as expected, all the AUC values are lower than the internal validations, ranging from 0.642 to 0.750 (Table \begin_inset CommandInset ref LatexCommand ref reference "tab:AUC-PAM" plural "false" caps "false" noprefix "false" \end_inset ). With or without GRSN, RMA shows its dominance over dChip in this more challengi ng test. Unlike in the internal validation, GRSN actually improves the classifier performance for RMA, although it does not for dChip. Once again, both single-channel methods perform about on par with RMA, with fRMA performing slightly better and SCAN performing a bit worse. Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:ROC-PAM-ext" plural "false" caps "false" noprefix "false" \end_inset shows the ROC curves for the external validation test. As expected, none of them are as clean-looking as the internal validation ROC curves. The curves for RMA, RMA+GRSN, and fRMA all look similar, while the other curves look more divergent. \end_layout \begin_layout Subsection fRMA with custom-generated vectors enables single-channel normalization on hthgu133pluspm platform \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Float figure placement tb wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/frma-pax-bx/batchsize_batches.pdf lyxscale 50 height 35theight% groupId frmatools-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:batch-size-batches" \end_inset \series bold Number of batches usable in fRMA probe weight learning as a function of batch size. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \begin_inset Float figure placement tb wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/frma-pax-bx/batchsize_samples.pdf lyxscale 50 height 35theight% groupId frmatools-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:batch-size-samples" \end_inset \series bold Number of samples usable in fRMA probe weight learning as a function of batch size. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:frmatools-batch-size" \end_inset Effect of batch size selection on number of batches and number of samples included in fRMA probe weight learning. \series default For batch sizes ranging from 3 to 15, the number of batches (a) and samples (b) included in probe weight training were plotted for biopsy (BX) and blood (PAX) samples. The selected batch size, 5, is marked with a dotted vertical line. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard In order to enable use of fRMA to normalize hthgu133pluspm, a custom set of fRMA vectors was created. First, an appropriate batch size was chosen by looking at the number of batches and number of samples included as a function of batch size (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:frmatools-batch-size" plural "false" caps "false" noprefix "false" \end_inset ). For a given batch size, all batches with fewer samples that the chosen size must be ignored during training, while larger batches must be randomly downsampled to the chosen size. Hence, the number of samples included for a given batch size equals the batch size times the number of batches with at least that many samples. From Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:batch-size-samples" plural "false" caps "false" noprefix "false" \end_inset , it is apparent that that a batch size of 8 maximizes the number of samples included in training. Increasing the batch size beyond this causes too many smaller batches to be excluded, reducing the total number of samples for both tissue types. However, a batch size of 8 is not necessarily optimal. The article introducing frmaTools concluded that it was highly advantageous to use a smaller batch size in order to include more batches, even at the expense of including fewer total samples in training \begin_inset CommandInset citation LatexCommand cite key "McCall2011" literal "false" \end_inset . To strike an appropriate balance between more batches and more samples, a batch size of 5 was chosen. For both blood and biopsy samples, this increased the number of batches included by 10, with only a modest reduction in the number of samples compared to a batch size of 8. With a batch size of 5, 26 batches of biopsy samples and 46 batches of blood samples were available. \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/frma-pax-bx/M-BX-violin.pdf lyxscale 40 width 45col% groupId m-violin \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:m-bx-violin" \end_inset \series bold Violin plot of inter-normalization log ratios for biopsy samples. \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/frma-pax-bx/M-PAX-violin.pdf lyxscale 40 width 45col% groupId m-violin \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:m-pax-violin" \end_inset \series bold Violin plot of inter-normalization log ratios for blood samples. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:frma-violin" \end_inset \series bold Violin plot of log ratios between normalizations for 20 biopsy samples. \series default Each of 20 randomly selected samples was normalized with RMA and with 5 different sets of fRMA vectors. The distribution of log ratios between normalized expression values, aggregated across all 20 arrays, was plotted for each pair of normalizations. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard Since fRMA training requires equal-size batches, larger batches are downsampled randomly. This introduces a nondeterministic step in the generation of normalization vectors. To show that this randomness does not substantially change the outcome, the random downsampling and subsequent vector learning was repeated 5 times, with a different random seed each time. 20 samples were selected at random as a test set and normalized with each of the 5 sets of fRMA normalization vectors as well as ordinary RMA, and the normalized expression values were compared across normalizations. Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:m-bx-violin" plural "false" caps "false" noprefix "false" \end_inset shows a summary of these comparisons for biopsy samples. Comparing RMA to each of the 5 fRMA normalizations, the distribution of log ratios is somewhat wide, indicating that the normalizations disagree on the expression values of a fair number of probe sets. In contrast, comparisons of fRMA against fRMA, the vast majority of probe sets have very small log ratios, indicating a very high agreement between the normalized values generated by the two normalizations. This shows that the fRMA normalization's behavior is not very sensitive to the random downsampling of larger batches during training. \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/frma-pax-bx/MA-BX-RMA.fRMA-RASTER.png lyxscale 10 width 45col% groupId ma-frma \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:ma-bx-rma-frma" \end_inset RMA vs. fRMA for biopsy samples. \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/frma-pax-bx/MA-BX-fRMA.fRMA-RASTER.png lyxscale 10 width 45col% groupId ma-frma \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:ma-bx-frma-frma" \end_inset fRMA vs fRMA for biopsy samples. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/frma-pax-bx/MA-PAX-RMA.fRMA-RASTER.png lyxscale 10 width 45col% groupId ma-frma \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:MA-PAX-rma-frma" \end_inset RMA vs. fRMA for blood samples. \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/frma-pax-bx/MA-PAX-fRMA.fRMA-RASTER.png lyxscale 10 width 45col% groupId ma-frma \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:MA-PAX-frma-frma" \end_inset fRMA vs fRMA for blood samples. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:Representative-MA-plots" \end_inset Representative MA plots comparing RMA and custom fRMA normalizations. \series default For each plot, 20 samples were normalized using 2 different normalizations, and then averages (A) and log ratios (M) were plotted between the two different normalizations for every probe. For the \begin_inset Quotes eld \end_inset fRMA vs fRMA \begin_inset Quotes erd \end_inset plots (b & d), two different fRMA normalizations using vectors from two independent batch samplings were compared. Density of points is represented by blue shading, and individual outlier points are plotted. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:ma-bx-rma-frma" plural "false" caps "false" noprefix "false" \end_inset shows an MA plot of the RMA-normalized values against the fRMA-normalized values for the same probe sets and arrays, corresponding to the first row of Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:m-bx-violin" plural "false" caps "false" noprefix "false" \end_inset . This MA plot shows that not only is there a wide distribution of M-values, but the trend of M-values is dependent on the average normalized intensity. This is expected, since the overall trend represents the differences in the quantile normalization step. When running RMA, only the quantiles for these specific 20 arrays are used, while for fRMA the quantile distribution is taking from all arrays used in training. Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:ma-bx-frma-frma" plural "false" caps "false" noprefix "false" \end_inset shows a similar MA plot comparing 2 different fRMA normalizations, correspondin g to the 6th row of Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:m-bx-violin" plural "false" caps "false" noprefix "false" \end_inset . The MA plot is very tightly centered around zero with no visible trend. Figures \begin_inset CommandInset ref LatexCommand ref reference "fig:m-pax-violin" plural "false" caps "false" noprefix "false" \end_inset , \begin_inset CommandInset ref LatexCommand ref reference "fig:MA-PAX-rma-frma" plural "false" caps "false" noprefix "false" \end_inset , and \begin_inset CommandInset ref LatexCommand ref reference "fig:ma-bx-frma-frma" plural "false" caps "false" noprefix "false" \end_inset show exactly the same information for the blood samples, once again comparing the normalized expression values between normalizations for all probe sets across 20 randomly selected test arrays. Once again, there is a wider distribution of log ratios between RMA-normalized values and fRMA-normalized, and a much tighter distribution when comparing different fRMA normalizations to each other, indicating that the fRMA training process is robust to random batch downsampling for the blood samples as well. \end_layout \begin_layout Subsection SVA, voom, and array weights improve model fit for methylation array data \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash afterpage{ \end_layout \begin_layout Plain Layout \backslash begin{landscape} \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Fix axis labels: \begin_inset Quotes eld \end_inset log2 M-value \begin_inset Quotes erd \end_inset is redundant because M-values are already log scale \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/methylvoom/unadj.dupcor/meanvar-trends-PAGE1-CROP-RASTER.png lyxscale 15 width 30col% groupId voomaw-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:meanvar-basic" \end_inset Mean-variance trend for analysis A. \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/methylvoom/unadj.dupcor.sva.aw/meanvar-trends-PAGE1-CROP-RASTER.png lyxscale 15 width 30col% groupId voomaw-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:meanvar-sva-aw" \end_inset Mean-variance trend for analysis B. \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/methylvoom/unadj.dupcor.sva.voomaw/meanvar-trends-PAGE2-CROP-RASTER.png lyxscale 15 width 30col% groupId voomaw-subfig \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:meanvar-sva-voomaw" \end_inset Mean-variance trend after voom modeling in analysis C. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold Mean-variance trend modeling in methylation array data. \series default The estimated log2(standard deviation) for each probe is plotted against the probe's average M-value across all samples as a black point, with some transparency to make over-plotting more visible, since there are about 450,000 points. Density of points is also indicated by the dark blue contour lines. The prior variance trend estimated by eBayes is shown in light blue, while the lowess trend of the points is shown in red. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash end{landscape} \end_layout \begin_layout Plain Layout } \end_layout \end_inset \end_layout \begin_layout Standard Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:meanvar-basic" plural "false" caps "false" noprefix "false" \end_inset shows the relationship between the mean M-value and the standard deviation calculated for each probe in the methylation array data set. A few features of the data are apparent. First, the data are very strongly bimodal, with peaks in the density around M-values of +4 and -4. These modes correspond to methylation sites that are nearly 100% methylated and nearly 100% unmethylated, respectively. The strong bimodality indicates that a majority of probes interrogate sites that fall into one of these two categories. The points in between these modes represent sites that are either partially methylated in many samples, or are fully methylated in some samples and fully unmethylated in other samples, or some combination. The next visible feature of the data is the W-shaped variance trend. The upticks in the variance trend on either side are expected, based on the sigmoid transformation exaggerating small differences at extreme M-values (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:Sigmoid-beta-m-mapping" plural "false" caps "false" noprefix "false" \end_inset ). However, the uptick in the center is interesting: it indicates that sites that are not constitutively methylated or unmethylated have a higher variance. This could be a genuine biological effect, or it could be spurious noise that is only observable at sites with varying methylation. \end_layout \begin_layout Standard In Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:meanvar-sva-aw" plural "false" caps "false" noprefix "false" \end_inset , we see the mean-variance trend for the same methylation array data, this time with surrogate variables and sample quality weights estimated from the data and included in the model. As expected, the overall average variance is smaller, since the surrogate variables account for some of the variance. In addition, the uptick in variance in the middle of the M-value range has disappeared, turning the W shape into a wide U shape. This indicates that the excess variance in the probes with intermediate M-values was explained by systematic variations not correlated with known covariates, and these variations were modeled by the surrogate variables. The result is a nearly flat variance trend for the entire intermediate M-value range from about -3 to +3. Note that this corresponds closely to the range within which the M-value transformation shown in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:Sigmoid-beta-m-mapping" plural "false" caps "false" noprefix "false" \end_inset is nearly linear. In contrast, the excess variance at the extremes (greater than +3 and less than -3) was not \begin_inset Quotes eld \end_inset absorbed \begin_inset Quotes erd \end_inset by the surrogate variables and remains in the plot, indicating that this variation has no systematic component: probes with extreme M-values are uniformly more variable across all samples, as expected. \end_layout \begin_layout Standard Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:meanvar-sva-voomaw" plural "false" caps "false" noprefix "false" \end_inset shows the mean-variance trend after fitting the model with the observation weights assigned by voom based on the mean-variance trend shown in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:meanvar-sva-aw" plural "false" caps "false" noprefix "false" \end_inset . As expected, the weights exactly counteract the trend in the data, resulting in a nearly flat trend centered vertically at 1 (i.e. 0 on the log scale). This shows that the observations with extreme M-values have been appropriately down-weighted to account for the fact that the noise in those observations has been amplified by the non-linear M-value transformation. In turn, this gives relatively more weight to observations in the middle region, which are more likely to correspond to probes measuring interesting biology (not constitutively methylated or unmethylated). \end_layout \begin_layout Standard \begin_inset Float table wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Tabular \begin_inset Text \begin_layout Plain Layout Covariate \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Test used \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout p-value \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Transplant Status \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout F-test \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0.404 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Diabetes Diagnosis \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \emph on t \emph default -test \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0.00106 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Sex \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \emph on t \emph default -test \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0.148 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Age \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout linear regression \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0.212 \end_layout \end_inset \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "tab:weight-covariate-tests" \end_inset Association of sample weights with clinical covariates in methylation array data. \series default Computed sample quality log weights were tested for significant association with each of the variables in the model (1st column). An appropriate test was selected for each variable based on whether the variable had 2 categories ( \emph on t \emph default -test), had more than 2 categories (F-test), or was numeric (linear regression). The test selected is shown in the 2nd column. P-values for association with the log weights are shown in the 3rd column. No multiple testing adjustment was performed for these p-values. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Redo the sample weight boxplot with notches, and remove fill colors \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/methylvoom/unadj.dupcor.sva.voomaw/sample-weights-PAGE3-CROP.pdf lyxscale 50 width 60col% groupId colwidth \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "fig:diabetes-sample-weights" \end_inset \series bold Box-and-whiskers plot of sample quality weights grouped by diabetes diagnosis. \series default Samples were grouped based on diabetes diagnosis, and the distribution of sample quality weights for each diagnosis was plotted as a box-and-whiskers plot \begin_inset CommandInset citation LatexCommand cite key "McGill1978" literal "false" \end_inset . \end_layout \end_inset \end_layout \begin_layout Plain Layout \end_layout \end_inset \end_layout \begin_layout Standard To determine whether any of the known experimental factors had an impact on data quality, the sample quality weights estimated from the data were tested for association with each of the experimental factors (Table \begin_inset CommandInset ref LatexCommand ref reference "tab:weight-covariate-tests" plural "false" caps "false" noprefix "false" \end_inset ). Diabetes diagnosis was found to have a potentially significant association with the sample weights, with a t-test p-value of \begin_inset Formula $1.06\times10^{-3}$ \end_inset . Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:diabetes-sample-weights" plural "false" caps "false" noprefix "false" \end_inset shows the distribution of sample weights grouped by diabetes diagnosis. The samples from patients with Type 2 diabetes were assigned significantly lower weights than those from patients with Type 1 diabetes. This indicates that the type 2 diabetes samples had an overall higher variance on average across all probes. \end_layout \begin_layout Standard \begin_inset Float table wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Consider transposing these tables \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Float table wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Tabular \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Analysis \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Contrast \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout A \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout B \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout C \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout TX vs AR \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 25 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 22 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout TX vs ADNR \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 7 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 338 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 369 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout TX vs CAN \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 231 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 278 \end_layout \end_inset \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "tab:methyl-num-signif" \end_inset Number of probes significant at 10% FDR. \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float table wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Tabular \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Analysis \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout Contrast \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout A \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout B \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout C \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout TX vs AR \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 0 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 10,063 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 11,225 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout TX vs ADNR \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 27 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 12,674 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 13,086 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout TX vs CAN \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 966 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 20,039 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout 20,955 \end_layout \end_inset \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset CommandInset label LatexCommand label name "tab:methyl-est-nonnull" \end_inset Estimated number of non-null tests, using the method of averaging local FDR values \begin_inset CommandInset citation LatexCommand cite key "Phipson2013Thesis" literal "false" \end_inset . \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold Estimates of degree of differential methylation in for each contrast in each analysis. \series default For each of the analyses in Table \begin_inset CommandInset ref LatexCommand ref reference "tab:Summary-of-meth-analysis" plural "false" caps "false" noprefix "false" \end_inset , these tables show the number of probes called significantly differentially methylated at a threshold of 10% FDR for each comparison between TX and the other 3 transplant statuses (a) and the estimated total number of probes that are differentially methylated (b). \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status open \begin_layout Plain Layout \align center \series bold \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/methylvoom/unadj.dupcor/pval-histograms-PAGE1.pdf lyxscale 33 width 30col% groupId meth-pval-hist \end_inset \end_layout \begin_layout Plain Layout \series bold \begin_inset Caption Standard \begin_layout Plain Layout AR vs. TX, Analysis A \end_layout \end_inset \end_layout \begin_layout Plain Layout \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/methylvoom/unadj.dupcor/pval-histograms-PAGE2.pdf lyxscale 33 width 30col% groupId meth-pval-hist \end_inset \end_layout \begin_layout Plain Layout \series bold \begin_inset Caption Standard \begin_layout Plain Layout ADNR vs. TX, Analysis A \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/methylvoom/unadj.dupcor/pval-histograms-PAGE3.pdf lyxscale 33 width 30col% groupId meth-pval-hist \end_inset \end_layout \begin_layout Plain Layout \series bold \begin_inset Caption Standard \begin_layout Plain Layout CAN vs. TX, Analysis A \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \series bold \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/methylvoom/unadj.dupcor.sva.aw/pval-histograms-PAGE1.pdf lyxscale 33 width 30col% groupId meth-pval-hist \end_inset \end_layout \begin_layout Plain Layout \series bold \begin_inset Caption Standard \begin_layout Plain Layout AR vs. TX, Analysis B \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/methylvoom/unadj.dupcor.sva.aw/pval-histograms-PAGE2.pdf lyxscale 33 width 30col% groupId meth-pval-hist \end_inset \end_layout \begin_layout Plain Layout \series bold \begin_inset Caption Standard \begin_layout Plain Layout ADNR vs. TX, Analysis B \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/methylvoom/unadj.dupcor.sva.aw/pval-histograms-PAGE3.pdf lyxscale 33 width 30col% groupId meth-pval-hist \end_inset \end_layout \begin_layout Plain Layout \series bold \begin_inset Caption Standard \begin_layout Plain Layout CAN vs. TX, Analysis B \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \align center \series bold \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/methylvoom/unadj.dupcor.sva.voomaw/pval-histograms-PAGE1.pdf lyxscale 33 width 30col% groupId meth-pval-hist \end_inset \end_layout \begin_layout Plain Layout \series bold \begin_inset Caption Standard \begin_layout Plain Layout AR vs. TX, Analysis C \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/methylvoom/unadj.dupcor.sva.voomaw/pval-histograms-PAGE2.pdf lyxscale 33 width 30col% groupId meth-pval-hist \end_inset \end_layout \begin_layout Plain Layout \series bold \begin_inset Caption Standard \begin_layout Plain Layout ADNR vs. TX, Analysis C \end_layout \end_inset \end_layout \end_inset \begin_inset space \hfill{} \end_inset \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/methylvoom/unadj.dupcor.sva.voomaw/pval-histograms-PAGE3.pdf lyxscale 33 width 30col% groupId meth-pval-hist \end_inset \end_layout \begin_layout Plain Layout \series bold \begin_inset Caption Standard \begin_layout Plain Layout CAN vs. TX, Analysis C \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset CommandInset label LatexCommand label name "fig:meth-p-value-histograms" \end_inset Probe p-value histograms for each contrast in each analysis. \series default For each differential methylation test of interest, the distribution of p-values across all probes is plotted as a histogram. The red solid line indicates the density that would be expected under the null hypothesis for all probes (a \begin_inset Formula $\mathrm{Uniform}(0,1)$ \end_inset distribution), while the blue dotted line indicates the fraction of p-values that actually follow the null hypothesis ( \begin_inset Formula $\hat{\pi}_{0}$ \end_inset ) estimated using the method of averaging local FDR values \begin_inset CommandInset citation LatexCommand cite key "Phipson2013Thesis" literal "false" \end_inset . the blue line is only shown in each plot if the estimate of \begin_inset Formula $\hat{\pi}_{0}$ \end_inset for that p-value distribution is different from 1. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard Table \begin_inset CommandInset ref LatexCommand ref reference "tab:methyl-num-signif" plural "false" caps "false" noprefix "false" \end_inset shows the number of significantly differentially methylated probes reported by each analysis for each comparison of interest at an FDR of 10%. As expected, the more elaborate analyses, B and C, report more significant probes than the more basic analysis A, consistent with the conclusions above that the data contain hidden systematic variations that must be modeled. Table \begin_inset CommandInset ref LatexCommand ref reference "tab:methyl-est-nonnull" plural "false" caps "false" noprefix "false" \end_inset shows the estimated number differentially methylated probes for each test from each analysis. This was computed by estimating the proportion of null hypotheses that were true using the method of \begin_inset CommandInset citation LatexCommand cite key "Phipson2013Thesis" literal "false" \end_inset and subtracting that fraction from the total number of probes, yielding an estimate of the number of null hypotheses that are false based on the distribution of p-values across the entire dataset. Note that this does not identify which null hypotheses should be rejected (i.e. which probes are significant); it only estimates the true number of such probes. Once again, analyses B and C result it much larger estimates for the number of differentially methylated probes. In this case, analysis C, the only analysis that includes voom, estimates the largest number of differentially methylated probes for all 3 contrasts. If the assumptions of all the methods employed hold, then this represents a gain in statistical power over the simpler analysis A. Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:meth-p-value-histograms" plural "false" caps "false" noprefix "false" \end_inset shows the p-value distributions for each test, from which the numbers in Table \begin_inset CommandInset ref LatexCommand ref reference "tab:methyl-est-nonnull" plural "false" caps "false" noprefix "false" \end_inset were generated. The distributions for analysis A all have a dip in density near zero, which is a strong sign of a poor model fit. The histograms for analyses B and C are more well-behaved, with a uniform component stretching all the way from 0 to 1 representing the probes for which the null hypotheses is true (no differential methylation), and a zero-biased component representing the probes for which the null hypothesis is false (differentially methylated). These histograms do not indicate any major issues with the model fit. \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout If time allows, maybe generate the PCA plots before/after SVA effect subtraction ? \end_layout \end_inset \end_layout \begin_layout Section Discussion \end_layout \begin_layout Subsection fRMA achieves clinically applicable normalization without sacrificing classifica tion performance \end_layout \begin_layout Standard As shown in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:Classifier-probabilities-RMA" plural "false" caps "false" noprefix "false" \end_inset , improper normalization, particularly separate normalization of training and test samples, leads to unwanted biases in classification. In a controlled experimental context, it is always possible to correct this issue by normalizing all experimental samples together. However, because it is not feasible to normalize all samples together in a clinical context, a single-channel normalization is required is required. \end_layout \begin_layout Standard The major concern in using a single-channel normalization is that non-single-cha nnel methods can share information between arrays to improve the normalization, and single-channel methods risk sacrificing the gains in normalization accuracy that come from this information sharing. In the case of RMA, this information sharing is accomplished through quantile normalization and median polish steps. The need for information sharing in quantile normalization can easily be removed by learning a fixed set of quantiles from external data and normalizing each array to these fixed quantiles, instead of the quantiles of the data itself. As long as the fixed quantiles are reasonable, the result will be similar to standard RMA. However, there is no analogous way to eliminate cross-array information sharing in the median polish step, so fRMA replaces this with a weighted average of probes on each array, with the weights learned from external data. This step of fRMA has the greatest potential to diverge from RMA un undesirable ways. \end_layout \begin_layout Standard However, when run on real data, fRMA performed at least as well as RMA in both the internal validation and external validation tests. This shows that fRMA can be used to normalize individual clinical samples in a class prediction context without sacrificing the classifier performance that would be obtained by using the more well-established RMA for normalization. The other single-channel normalization method considered, SCAN, showed some loss of AUC in the external validation test. Based on these results, fRMA is the preferred normalization for clinical samples in a class prediction context. \end_layout \begin_layout Subsection Robust fRMA vectors can be generated for new array platforms \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Look up the exact numbers, do a find & replace for \begin_inset Quotes eld \end_inset 850 \begin_inset Quotes erd \end_inset \end_layout \end_inset \end_layout \begin_layout Standard The published fRMA normalization vectors for the hgu133plus2 platform were generated from a set of about 850 samples chosen from a wide range of tissues, which the authors determined was sufficient to generate a robust set of normalization vectors that could be applied across all tissues \begin_inset CommandInset citation LatexCommand cite key "McCall2010" literal "false" \end_inset . Since we only had hthgu133pluspm for 2 tissues of interest, our needs were more modest. Even using only 130 samples in 26 batches of 5 samples each for kidney biopsies, we were able to train a robust set of fRMA normalization vectors that were not meaningfully affected by the random selection of 5 samples from each batch. As expected, the training process was just as robust for the blood samples with 230 samples in 46 batches of 5 samples each. Because these vectors were each generated using training samples from a single tissue, they are not suitable for general use, unlike the vectors provided with fRMA itself. They are purpose-built for normalizing a specific type of sample on a specific platform. This is a mostly acceptable limitation in the context of developing a machine learning classifier for diagnosing a disease based on samples of a specific tissue. \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Talk about how these vectors can be used for any data from these tissues on this platform even though they were custom made for this data set. \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout How to bring up that these custom vectors were used in another project by someone else that was never published? \end_layout \end_inset \end_layout \begin_layout Subsection Methylation array data can be successfully analyzed using existing techniques, but machine learning poses additional challenges \end_layout \begin_layout Standard Both analysis strategies B and C both yield a reasonable analysis, with a mean-variance trend that matches the expected behavior for the non-linear M-value transformation (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:meanvar-sva-aw" plural "false" caps "false" noprefix "false" \end_inset ) and well-behaved p-value distributions (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:meth-p-value-histograms" plural "false" caps "false" noprefix "false" \end_inset ). These two analyses also yield similar numbers of significant probes (Table \begin_inset CommandInset ref LatexCommand ref reference "tab:methyl-num-signif" plural "false" caps "false" noprefix "false" \end_inset ) and similar estimates of the number of differentially methylated probes (Table \begin_inset CommandInset ref LatexCommand ref reference "tab:methyl-est-nonnull" plural "false" caps "false" noprefix "false" \end_inset ). The main difference between these two analyses is the method used to account for the mean-variance trend. In analysis B, the trend is estimated and applied at the probe level: each probe's estimated variance is squeezed toward the trend using an empirical Bayes procedure (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:meanvar-sva-aw" plural "false" caps "false" noprefix "false" \end_inset ). In analysis C, the trend is still estimated at the probe level, but instead of estimating a single variance value shared across all observations for a given probe, the voom method computes an initial estimate of the variance for each observation individually based on where its model-fitted M-value falls on the trend line and then assigns inverse-variance weights to model the difference in variance between observations. An overall variance is still estimated for each probe using the same empirical Bayes method, but now the residual trend is flat (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:meanvar-sva-voomaw" plural "false" caps "false" noprefix "false" \end_inset ), indicating that the mean-variance trend is adequately modeled by scaling the estimated variance for each observation using the weights computed by voom. \end_layout \begin_layout Standard The difference between the standard empirical Bayes trended variance modeling (analysis B) and voom (analysis C) is analogous to the difference between a t-test with equal variance and a t-test with unequal variance, except that the unequal group variances used in the latter test are estimated based on the mean-variance trend from all the probes rather than the data for the specific probe being tested, thus stabilizing the group variance estimates by sharing information between probes. Allowing voom to model the variance using observation weights in this manner allows the linear model fit to concentrate statistical power where it will do the most good. For example, if a particular probe's M-values are always at the extreme of the M-value range (e.g. less than -4) for ADNR samples, but the M-values for that probe in TX and CAN samples are within the flat region of the mean-variance trend (between -3 and +3), voom is able to down-weight the contribution of the high-variance M-values from the ADNR samples in order to gain more statistical power while testing for differential methylation between TX and CAN. In contrast, modeling the mean-variance trend only at the probe level would combine the high-variance ADNR samples and lower-variance samples from other conditions and estimate an intermediate variance for this probe. In practice, analysis B shows that this approach is adequate, but the voom approach in analysis C is at least as good on all model fit criteria and yields a larger estimate for the number of differentially methylated genes, \emph on and \emph default it matches up better with the theoretical \end_layout \begin_layout Standard The significant association of diabetes diagnosis with sample quality is interesting. The samples with Type 2 diabetes tended to have more variation, averaged across all probes, than those with Type 1 diabetes. This is consistent with the consensus that type 2 diabetes and the associated metabolic syndrome represent a broad dysregulation of the body's endocrine signaling related to metabolism [citation needed]. This dysregulation could easily manifest as a greater degree of variation in the DNA methylation patterns of affected tissues. In contrast, Type 1 diabetes has a more specific cause and effect, so a less variable methylation signature is expected. \end_layout \begin_layout Standard This preliminary analysis suggests that some degree of differential methylation exists between TX and each of the three types of transplant disfunction studied. Hence, it may be feasible to train a classifier to diagnose transplant disfunction from DNA methylation array data. However, the major importance of both SVA and sample quality weighting for proper modeling of this data poses significant challenges for any attempt at a machine learning on data of similar quality. While these are easily used in a modeling context with full sample information, neither of these methods is directly applicable in a machine learning context, where the diagnosis is not known ahead of time. If a machine learning approach for methylation-based diagnosis is to be pursued, it will either require machine-learning-friendly methods to address the same systematic trends in the data that SVA and sample quality weighting address, or it will require higher quality data with substantially less systematic perturbation of the data. \end_layout \begin_layout Section Future Directions \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Some work was already being done with the existing fRMA vectors. Do I mention that here? \end_layout \end_inset \end_layout \begin_layout Subsection Improving fRMA to allow training from batches of unequal size \end_layout \begin_layout Standard Because the tools for building fRMA normalization vectors require equal-size batches, many samples must be discarded from the training data. This is undesirable for a few reasons. First, more data is simply better, all other things being equal. In this case, \begin_inset Quotes eld \end_inset better \begin_inset Quotes erd \end_inset means a more precise estimate of normalization parameters. In addition, the samples to be discarded must be chosen arbitrarily, which introduces an unnecessary element of randomness into the estimation process. While the randomness can be made deterministic by setting a consistent random seed, the need for equal size batches also introduces a need for the analyst to decide on the appropriate trade-off between batch size and the number of batches. This introduces an unnecessary and undesirable \begin_inset Quotes eld \end_inset researcher degree of freedom \begin_inset Quotes erd \end_inset into the analysis, since the generated normalization vectors now depend on the choice of batch size based on vague selection criteria and instinct, which can unintentionally introduce bias if the researcher chooses a batch size based on what seems to yield the most favorable downstream results \begin_inset CommandInset citation LatexCommand cite key "Simmons2011" literal "false" \end_inset . \end_layout \begin_layout Standard Fortunately, the requirement for equal-size batches is not inherent to the fRMA algorithm but rather a limitation of the implementation in the frmaTools package. In personal communication, the package's author, Matthew McCall, has indicated that with some work, it should be possible to improve the implementation to work with batches of unequal sizes. The current implementation ignores the batch size when calculating with-batch and between-batch residual variances, since the batch size constant cancels out later in the calculations as long as all batches are of equal size. Hence, the calculations of these parameters would need to be modified to remove this optimization and properly calculate the variances using the full formula. Once this modification is made, a new strategy would need to be developed for assessing the stability of parameter estimates, since the random subsamplin g step is eliminated, meaning that different subsamplings can no longer be compared as in Figures \begin_inset CommandInset ref LatexCommand ref reference "fig:frma-violin" plural "false" caps "false" noprefix "false" \end_inset and \begin_inset CommandInset ref LatexCommand ref reference "fig:Representative-MA-plots" plural "false" caps "false" noprefix "false" \end_inset . Bootstrap resampling is likely a good candidate here: sample many training sets of equal size from the existing training set with replacement, estimate parameters from each resampled training set, and compare the estimated parameters between bootstraps in order to quantify the variability in each parameter's estimation. \end_layout \begin_layout Subsection Developing methylation arrays as a diagnostic tool for kidney transplant rejection \end_layout \begin_layout Standard The current study has showed that DNA methylation, as assayed by Illumina 450k methylation arrays, has some potential for diagnosing transplant dysfuncti ons, including rejection. However, very few probes could be confidently identified as differentially methylated between healthy and dysfunctional transplants. One likely explanation for this is the predominant influence of unobserved confounding factors. SVA can model and correct for such factors, but the correction can never be perfect, so some degree of unwanted systematic variation will always remain after SVA correction. If the effect size of the confounding factors was similar to that of the factor of interest (in this case, transplant status), this would be an acceptable limitation, since removing most of the confounding factors' effects would allow the main effect to stand out. However, in this data set, the confounding factors have a much larger effect size than transplant status, which means that the small degree of remaining variation not removed by SVA can still swamp the effect of interest, making it difficult to detect. This is, of course, a major issue when the end goal is to develop a classifier to diagnose transplant rejection from methylation data, since batch-correction methods like SVA that work in a linear modeling context cannot be applied in a machine learning context. \end_layout \begin_layout Standard Currently, the source of these unwanted systematic variations in the data is unknown. The best solution would be to determine the cause of the variation and eliminate it, thereby eliminating the need to model and remove that variation. However, if this proves impractical, another option is to use SVA to identify probes that are highly associated with the surrogate variables that describe the unwanted variation in the data. These probes could be discarded prior to classifier training, in order to maximize the chance that the training algorithm will be able to identify highly predictive probes from those remaining. Lastly, it is possible that some of this unwanted variation is a result of the array-based assay being used and would be eliminated by switching to assaying DNA methylation using bisulphite sequencing. However, this carries the risk that the sequencing assay will have its own set of biases that must be corrected for in a different way. \end_layout \begin_layout Chapter Globin-blocking for more effective blood RNA-seq analysis in primate animal model \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Choose between above and the paper title: Optimizing yield of deep RNA sequencin g for gene expression profiling by globin reduction of peripheral blood samples from cynomolgus monkeys (Macaca fascicularis). \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Chapter author list: \begin_inset CommandInset href LatexCommand href target "https://tex.stackexchange.com/questions/156862/displaying-author-for-each-chapter-in-book" \end_inset Every chapter gets an author list, which may or may not be part of a citation to a published/preprinted paper. \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Preprint then cite the paper \end_layout \end_inset \end_layout \begin_layout Section* Abstract \end_layout \begin_layout Paragraph Background \end_layout \begin_layout Standard Primate blood contains high concentrations of globin messenger RNA. Globin reduction is a standard technique used to improve the expression results obtained by DNA microarrays on RNA from blood samples. However, with whole transcriptome RNA-sequencing (RNA-seq) quickly replacing microarrays for many applications, the impact of globin reduction for RNA-seq has not been previously studied. Moreover, no off-the-shelf kits are available for globin reduction in nonhuman primates. \end_layout \begin_layout Paragraph Results \end_layout \begin_layout Standard Here we report a protocol for RNA-seq in primate blood samples that uses complimentary oligonucleotides to block reverse transcription of the alpha and beta globin genes. In test samples from cynomolgus monkeys (Macaca fascicularis), this globin blocking protocol approximately doubles the yield of informative (non-globin) reads by greatly reducing the fraction of globin reads, while also improving the consistency in sequencing depth between samples. The increased yield enables detection of about 2000 more genes, significantly increases the correlation in measured gene expression levels between samples, and increases the sensitivity of differential gene expression tests. \end_layout \begin_layout Paragraph Conclusions \end_layout \begin_layout Standard These results show that globin blocking significantly improves the cost-effectiv eness of mRNA sequencing in primate blood samples by doubling the yield of useful reads, allowing detection of more genes, and improving the precision of gene expression measurements. Based on these results, a globin reducing or blocking protocol is recommended for all RNA-seq studies of primate blood samples. \end_layout \begin_layout Section Approach \end_layout \begin_layout Standard \begin_inset Note Note status open \begin_layout Plain Layout Consider putting some of this in the Intro chapter \end_layout \begin_layout Itemize Cynomolgus monkeys as a model organism \end_layout \begin_deeper \begin_layout Itemize Highly related to humans \end_layout \begin_layout Itemize Small size and short life cycle - good research animal \end_layout \begin_layout Itemize Genomics resources still in development \end_layout \end_deeper \begin_layout Itemize Inadequacy of existing blood RNA-seq protocols \end_layout \begin_deeper \begin_layout Itemize Existing protocols use a separate globin pulldown step, slowing down processing \end_layout \end_deeper \end_inset \end_layout \begin_layout Standard Increasingly, researchers are turning to high-throughput mRNA sequencing technologies (RNA-seq) in preference to expression microarrays for analysis of gene expression \begin_inset CommandInset citation LatexCommand cite key "Mutz2012" literal "false" \end_inset . The advantages are even greater for study of model organisms with no well-estab lished array platforms available, such as the cynomolgus monkey (Macaca fascicularis). High fractions of globin mRNA are naturally present in mammalian peripheral blood samples (up to 70% of total mRNA) and these are known to interfere with the results of array-based expression profiling \begin_inset CommandInset citation LatexCommand cite key "Winn2010" literal "false" \end_inset . The importance of globin reduction for RNA-seq of blood has only been evaluated for a deepSAGE protocol on human samples \begin_inset CommandInset citation LatexCommand cite key "Mastrokolias2012" literal "false" \end_inset . In the present report, we evaluated globin reduction using custom blocking oligonucleotides for deep RNA-seq of peripheral blood samples from a nonhuman primate, cynomolgus monkey, using the Illumina technology platform. We demonstrate that globin reduction significantly improves the cost-effectiven ess of RNA-seq in blood samples. Thus, our protocol offers a significant advantage to any investigator planning to use RNA-seq for gene expression profiling of nonhuman primate blood samples. Our method can be generally applied to any species by designing complementary oligonucleotide blocking probes to the globin gene sequences of that species. Indeed, any highly expressed but biologically uninformative transcripts can also be blocked to further increase sequencing efficiency and value \begin_inset CommandInset citation LatexCommand cite key "Arnaud2016" literal "false" \end_inset . \end_layout \begin_layout Section Methods \end_layout \begin_layout Subsection Sample collection \end_layout \begin_layout Standard All research reported here was done under IACUC-approved protocols at the University of Miami and complied with all applicable federal and state regulations and ethical principles for nonhuman primate research. Blood draws occurred between 16 April 2012 and 18 June 2015. The experimental system involved intrahepatic pancreatic islet transplantation into Cynomolgus monkeys with induced diabetes mellitus with or without concomitant infusion of mesenchymal stem cells. Blood was collected at serial time points before and after transplantation into PAXgene Blood RNA tubes (PreAnalytiX/Qiagen, Valencia, CA) at the precise volume:volume ratio of 2.5 ml whole blood into 6.9 ml of PAX gene additive. \end_layout \begin_layout Subsection Globin Blocking \end_layout \begin_layout Standard Four oligonucleotides were designed to hybridize to the 3’ end of the transcript s for Cynomolgus HBA1, HBA2 and HBB, with two hybridization sites for HBB and 2 sites for HBA (the chosen sites were identical in both HBA genes). All oligos were purchased from Sigma and were entirely composed of 2’O-Me bases with a C3 spacer positioned at the 3’ ends to prevent any polymerase mediated primer extension. \end_layout \begin_layout Quote HBA1/2 site 1: GCCCACUCAGACUUUAUUCAAAG-C3spacer \end_layout \begin_layout Quote HBA1/2 site 2: GGUGCAAGGAGGGGAGGAG-C3spacer \end_layout \begin_layout Quote HBB site 1: AAUGAAAAUAAAUGUUUUUUAUUAG-C3spacer \end_layout \begin_layout Quote HBB site 2: CUCAAGGCCCUUCAUAAUAUCCC-C3spacer \end_layout \begin_layout Subsection RNA-seq Library Preparation \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Add protected spaces where appropriate to prevent unwanted line breaks. \end_layout \end_inset \end_layout \begin_layout Standard Sequencing libraries were prepared with 200 \begin_inset space ~ \end_inset ng total RNA from each sample. Polyadenylated mRNA was selected from 200 ng aliquots of cynomolgus blood-deriv ed total RNA using Ambion Dynabeads Oligo(dT)25 beads (Invitrogen) following manufacturer’s recommended protocol. PolyA selected RNA was then combined with 8 pmol of HBA1/2 (site 1), 8 pmol of HBA1/2 (site 2), 12 pmol of HBB (site 1) and 12 pmol of HBB (site 2) oligonucleotides. In addition, 20 pmol of RT primer containing a portion of the Illumina adapter sequence (B-oligo-dTV: GAGTTCCTTGGCACCCGAGAATTCCATTTTTTTTTTTTTTTTTTTV) and 4 µL of 5X First Strand buffer (250 mM Tris-HCl pH 8.3, 375 mM KCl, 15mM MgCl2) were added in a total volume of 15 µL. The RNA was fragmented by heating this cocktail for 3 minutes at 95°C and then placed on ice. This was followed by the addition of 2 µL 0.1 M DTT, 1 µL RNaseOUT, 1 µL 10mM dNTPs 10% biotin-16 aminoallyl-2’- dUTP and 10% biotin-16 aminoallyl-2’- dCTP (TriLink Biotech, San Diego, CA), 1 µL Superscript II (200U/ µL, Thermo-Fi sher). A second “unblocked” library was prepared in the same way for each sample but replacing the blocking oligos with an equivalent volume of water. The reaction was carried out at 25°C for 15 minutes and 42°C for 40 minutes, followed by incubation at 75°C for 10 minutes to inactivate the reverse transcriptase. \end_layout \begin_layout Standard The cDNA/RNA hybrid molecules were purified using 1.8X Ampure XP beads (Agencourt ) following supplier’s recommended protocol. The cDNA/RNA hybrid was eluted in 25 µL of 10 mM Tris-HCl pH 8.0, and then bound to 25 µL of M280 Magnetic Streptavidin beads washed per recommended protocol (Thermo-Fisher). After 30 minutes of binding, beads were washed one time in 100 µL 0.1N NaOH to denature and remove the bound RNA, followed by two 100 µL washes with 1X TE buffer. \end_layout \begin_layout Standard Subsequent attachment of the 5-prime Illumina A adapter was performed by on-bead random primer extension of the following sequence (A-N8 primer: TTCAGAGTTCTACAGTCCGACGATCNNNNNNNN). Briefly, beads were resuspended in a 20 µL reaction containing 5 µM A-N8 primer, 40mM Tris-HCl pH 7.5, 20mM MgCl2, 50mM NaCl, 0.325U/µL Sequenase 2.0 (Affymetrix, Santa Clara, CA), 0.0025U/µL inorganic pyrophosphatase (Affymetr ix) and 300 µM each dNTP. Reaction was incubated at 22°C for 30 minutes, then beads were washed 2 times with 1X TE buffer (200µL). \end_layout \begin_layout Standard The magnetic streptavidin beads were resuspended in 34 µL nuclease-free water and added directly to a PCR tube. The two Illumina protocol-specified PCR primers were added at 0.53 µM (Illumina TruSeq Universal Primer 1 and Illumina TruSeq barcoded PCR primer 2), along with 40 µL 2X KAPA HiFi Hotstart ReadyMix (KAPA, Willmington MA) and thermocycl ed as follows: starting with 98°C (2 min-hold); 15 cycles of 98°C, 20sec; 60°C, 30sec; 72°C, 30sec; and finished with a 72°C (2 min-hold). \end_layout \begin_layout Standard PCR products were purified with 1X Ampure Beads following manufacturer’s recommended protocol. Libraries were then analyzed using the Agilent TapeStation and quantitation of desired size range was performed by “smear analysis”. Samples were pooled in equimolar batches of 16 samples. Pooled libraries were size selected on 2% agarose gels (E-Gel EX Agarose Gels; Thermo-Fisher). Products were cut between 250 and 350 bp (corresponding to insert sizes of 130 to 230 bps). Finished library pools were then sequenced on the Illumina NextSeq500 instrumen t with 75 base read lengths. \end_layout \begin_layout Subsection Read alignment and counting \end_layout \begin_layout Standard Reads were aligned to the cynomolgus genome using STAR \begin_inset CommandInset citation LatexCommand cite key "Dobin2013,Wilson2013" literal "false" \end_inset . Counts of uniquely mapped reads were obtained for every gene in each sample with the \begin_inset Flex Code status open \begin_layout Plain Layout featureCounts \end_layout \end_inset function from the \begin_inset Flex Code status open \begin_layout Plain Layout Rsubread \end_layout \end_inset package, using each of the three possibilities for the \begin_inset Flex Code status open \begin_layout Plain Layout strandSpecific \end_layout \end_inset option: sense, antisense, and unstranded \begin_inset CommandInset citation LatexCommand cite key "Liao2014" literal "false" \end_inset . A few artifacts in the cynomolgus genome annotation complicated read counting. First, no ortholog is annotated for alpha globin in the cynomolgus genome, presumably because the human genome has two alpha globin genes with nearly identical sequences, making the orthology relationship ambiguous. However, two loci in the cynomolgus genome are as “hemoglobin subunit alpha-lik e” (LOC102136192 and LOC102136846). LOC102136192 is annotated as a pseudogene while LOC102136846 is annotated as protein-coding. Our globin reduction protocol was designed to include blocking of these two genes. Indeed, these two genes have almost the same read counts in each library as the properly-annotated HBB gene and much larger counts than any other gene in the unblocked libraries, giving confidence that reads derived from the real alpha globin are mapping to both genes. Thus, reads from both of these loci were counted as alpha globin reads in all further analyses. The second artifact is a small, uncharacterized non-coding RNA gene (LOC1021365 91), which overlaps the HBA-like gene (LOC102136192) on the opposite strand. If counting is not performed in stranded mode (or if a non-strand-specific sequencing protocol is used), many reads mapping to the globin gene will be discarded as ambiguous due to their overlap with this ncRNA gene, resulting in significant undercounting of globin reads. Therefore, stranded sense counts were used for all further analysis in the present study to insure that we accurately accounted for globin transcript reduction. However, we note that stranded reads are not necessary for RNA-seq using our protocol in standard practice. \end_layout \begin_layout Subsection Normalization and Exploratory Data Analysis \end_layout \begin_layout Standard Libraries were normalized by computing scaling factors using the \begin_inset Flex Code status open \begin_layout Plain Layout edgeR \end_layout \end_inset package’s Trimmed Mean of M-values method \begin_inset CommandInset citation LatexCommand cite key "Robinson2010" literal "false" \end_inset . Log2 counts per million values (logCPM) were calculated using the cpm function in \begin_inset Flex Code status open \begin_layout Plain Layout edgeR \end_layout \end_inset for individual samples and aveLogCPM function for averages across groups of samples, using those functions’ default prior count values to avoid taking the logarithm of 0. Genes were considered “present” if their average normalized logCPM values across all libraries were at least -1. Normalizing for gene length was unnecessary because the sequencing protocol is 3’-biased and hence the expected read count for each gene is related to the transcript’s copy number but not its length. \end_layout \begin_layout Standard In order to assess the effect of blocking on reproducibility, Pearson and Spearman correlation coefficients were computed between the logCPM values for every pair of libraries within the globin-blocked (GB) and unblocked (non-GB) groups, and \begin_inset Flex Code status open \begin_layout Plain Layout edgeR \end_layout \end_inset 's \begin_inset Flex Code status open \begin_layout Plain Layout estimateDisp \end_layout \end_inset function was used to compute negative binomial dispersions separately for the two groups \begin_inset CommandInset citation LatexCommand cite key "Chen2014" literal "false" \end_inset . \end_layout \begin_layout Subsection Differential Expression Analysis \end_layout \begin_layout Standard All tests for differential gene expression were performed using \begin_inset Flex Code status open \begin_layout Plain Layout edgeR \end_layout \end_inset , by first fitting a negative binomial generalized linear model to the counts and normalization factors and then performing a quasi-likelihood F-test with robust estimation of outlier gene dispersions \begin_inset CommandInset citation LatexCommand cite key "Lund2012,Phipson2016" literal "false" \end_inset . To investigate the effects of globin blocking on each gene, an additive model was fit to the full data with coefficients for globin blocking and SampleID. To test the effect of globin blocking on detection of differentially expressed genes, the GB samples and non-GB samples were each analyzed independently as follows: for each animal with both a pre-transplant and a post-transplant time point in the data set, the pre-transplant sample and the earliest post-transplant sample were selected, and all others were excluded, yielding a pre-/post-transplant pair of samples for each animal (N=7 animals with paired samples). These samples were analyzed for pre-transplant vs. post-transplant differential gene expression while controlling for inter-animal variation using an additive model with coefficients for transplant and animal ID. In all analyses, p-values were adjusted using the Benjamini-Hochberg procedure for FDR control \begin_inset CommandInset citation LatexCommand cite key "Benjamini1995" literal "false" \end_inset . \end_layout \begin_layout Standard \begin_inset Note Note status open \begin_layout Itemize New blood RNA-seq protocol to block reverse transcription of globin genes \end_layout \begin_layout Itemize Blood RNA-seq time course after transplants with/without MSC infusion \end_layout \end_inset \end_layout \begin_layout Section Results \end_layout \begin_layout Subsection Globin blocking yields a larger and more consistent fraction of useful reads \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash afterpage{ \end_layout \begin_layout Plain Layout \backslash begin{landscape} \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float table placement p wide false sideways false status open \begin_layout Plain Layout \align center \begin_inset Tabular \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none Percent of Total Reads \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none Percent of Genic Reads \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout GB \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none Non-globin Reads \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none Globin Reads \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none All Genic Reads \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none All Aligned Reads \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none Non-globin Reads \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none Globin Reads \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none Yes \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 50.4% ± 6.82 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 3.48% ± 2.94 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 53.9% ± 6.81 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 89.7% ± 2.40 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 93.5% ± 5.25 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 6.49% ± 5.25 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none No \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 26.3% ± 8.95 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 44.6% ± 16.6 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 70.1% ± 9.38 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 90.7% ± 5.16 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 38.8% ± 17.1 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 61.2% ± 17.1 \end_layout \end_inset \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset Argument 1 status collapsed \begin_layout Plain Layout Fractions of reads mapping to genomic features in GB and non-GB samples. \end_layout \end_inset \begin_inset CommandInset label LatexCommand label name "tab:Fractions-of-reads" \end_inset Fractions of reads mapping to genomic features in GB and non-GB samples. \series default All values are given as mean ± standard deviation. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset ERT status open \begin_layout Plain Layout \backslash end{landscape} \end_layout \begin_layout Plain Layout } \end_layout \end_inset \end_layout \begin_layout Standard The objective of the present study was to validate a new protocol for deep RNA-seq of whole blood drawn into PaxGene tubes from cynomolgus monkeys undergoing islet transplantation, with particular focus on minimizing the loss of useful sequencing space to uninformative globin reads. The details of the analysis with respect to transplant outcomes and the impact of mesenchymal stem cell treatment will be reported in a separate manuscript (in preparation). To focus on the efficacy of our globin blocking protocol, 37 blood samples, 16 from pre-transplant and 21 from post-transplant time points, were each prepped once with and once without globin blocking oligos, and were then sequenced on an Illumina NextSeq500 instrument. The number of reads aligning to each gene in the cynomolgus genome was counted. Table 1 summarizes the distribution of read fractions among the GB and non-GB libraries. In the libraries with no globin blocking, globin reads made up an average of 44.6% of total input reads, while reads assigned to all other genes made up an average of 26.3%. The remaining reads either aligned to intergenic regions (that include long non-coding RNAs) or did not align with any annotated transcripts in the current build of the cynomolgus genome. In the GB libraries, globin reads made up only 3.48% and reads assigned to all other genes increased to 50.4%. Thus, globin blocking resulted in a 92.2% reduction in globin reads and a 91.6% increase in yield of useful non-globin reads. \end_layout \begin_layout Standard This reduction is not quite as efficient as the previous analysis showed for human samples by DeepSAGE (<0.4% globin reads after globin reduction) \begin_inset CommandInset citation LatexCommand cite key "Mastrokolias2012" literal "false" \end_inset . Nonetheless, this degree of globin reduction is sufficient to nearly double the yield of useful reads. Thus, globin blocking cuts the required sequencing effort (and costs) to achieve a target coverage depth by almost 50%. Consistent with this near doubling of yield, the average difference in un-normalized logCPM across all genes between the GB libraries and non-GB libraries is approximately 1 (mean = 1.01, median = 1.08), an overall 2-fold increase. Un-normalized values are used here because the TMM normalization correctly identifies this 2-fold difference as biologically irrelevant and removes it. \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/Globin Paper/figure1 - globin-fractions.pdf lyxscale 50 width 75col% \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset Argument 1 status collapsed \begin_layout Plain Layout Fraction of genic reads in each sample aligned to non-globin genes, with and without globin blocking (GB). \end_layout \end_inset \begin_inset CommandInset label LatexCommand label name "fig:Fraction-of-genic-reads" \end_inset Fraction of genic reads in each sample aligned to non-globin genes, with and without globin blocking (GB). \series default All reads in each sequencing library were aligned to the cyno genome, and the number of reads uniquely aligning to each gene was counted. For each sample, counts were summed separately for all globin genes and for the remainder of the genes (non-globin genes), and the fraction of genic reads aligned to non-globin genes was computed. Each point represents an individual sample. Gray + signs indicate the means for globin-blocked libraries and unblocked libraries. The overall distribution for each group is represented as a notched box plots. Points are randomly spread vertically to avoid excessive overlapping. \end_layout \end_inset \end_layout \end_inset \end_layout \begin_layout Standard Another important aspect is that the standard deviations in Table \begin_inset CommandInset ref LatexCommand ref reference "tab:Fractions-of-reads" plural "false" caps "false" noprefix "false" \end_inset are uniformly smaller in the GB samples than the non-GB ones, indicating much greater consistency of yield. This is best seen in the percentage of non-globin reads as a fraction of total reads aligned to annotated genes (genic reads). For the non-GB samples, this measure ranges from 10.9% to 80.9%, while for the GB samples it ranges from 81.9% to 99.9% (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:Fraction-of-genic-reads" plural "false" caps "false" noprefix "false" \end_inset ). This means that for applications where it is critical that each sample achieve a specified minimum coverage in order to provide useful information, it would be necessary to budget up to 10 times the sequencing depth per sample without globin blocking, even though the average yield improvement for globin blocking is only 2-fold, because every sample has a chance of being 90% globin and 10% useful reads. Hence, the more consistent behavior of GB samples makes planning an experiment easier and more efficient because it eliminates the need to over-sequence every sample in order to guard against the worst case of a high-globin fraction. \end_layout \begin_layout Subsection Globin blocking lowers the noise floor and allows detection of about 2000 more low-expression genes \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Remove redundant titles from figures \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/Globin Paper/figure2 - aveLogCPM-colored.pdf lyxscale 50 height 60theight% \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset Argument 1 status collapsed \begin_layout Plain Layout Distributions of average group gene abundances when normalized separately or together. \end_layout \end_inset \begin_inset CommandInset label LatexCommand label name "fig:logcpm-dists" \end_inset Distributions of average group gene abundances when normalized separately or together. \series default All reads in each sequencing library were aligned to the cyno genome, and the number of reads uniquely aligning to each gene was counted. Genes with zero counts in all libraries were discarded. Libraries were normalized using the TMM method. Libraries were split into globin-blocked (GB) and non-GB groups and the average abundance for each gene in both groups, measured in log2 counts per million reads counted, was computed using the aveLogCPM function. The distribution of average gene logCPM values was plotted for both groups using a kernel density plot to approximate a continuous distribution. The logCPM GB distributions are marked in red, non-GB in blue. The black vertical line denotes the chosen detection threshold of -1. Top panel: Libraries were split into GB and non-GB groups first and normalized separately. Bottom panel: Libraries were all normalized together first and then split into groups. \end_layout \end_inset \end_layout \begin_layout Plain Layout \end_layout \end_inset \end_layout \begin_layout Standard Since globin blocking yields more usable sequencing depth, it should also allow detection of more genes at any given threshold. When we looked at the distribution of average normalized logCPM values across all libraries for genes with at least one read assigned to them, we observed the expected bimodal distribution, with a high-abundance "signal" peak representing detected genes and a low-abundance "noise" peak representing genes whose read count did not rise above the noise floor (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:logcpm-dists" plural "false" caps "false" noprefix "false" \end_inset ). Consistent with the 2-fold increase in raw counts assigned to non-globin genes, the signal peak for GB samples is shifted to the right relative to the non-GB signal peak. When all the samples are normalized together, this difference is normalized out, lining up the signal peaks, and this reveals that, as expected, the noise floor for the GB samples is about 2-fold lower. This greater separation between signal and noise peaks in the GB samples means that low-expression genes should be more easily detected and more precisely quantified than in the non-GB samples. \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/Globin Paper/figure3 - detection.pdf lyxscale 50 width 70col% \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset Argument 1 status collapsed \begin_layout Plain Layout Gene detections as a function of abundance thresholds in globin-blocked (GB) and non-GB samples. \end_layout \end_inset \begin_inset CommandInset label LatexCommand label name "fig:Gene-detections" \end_inset Gene detections as a function of abundance thresholds in globin-blocked (GB) and non-GB samples. \series default Average abundance (logCPM, \begin_inset Formula $\log_{2}$ \end_inset counts per million reads counted) was computed by separate group normalization as described in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:logcpm-dists" plural "false" caps "false" noprefix "false" \end_inset for both the GB and non-GB groups, as well as for all samples considered as one large group. For each every integer threshold from -2 to 3, the number of genes detected at or above that logCPM threshold was plotted for each group. \end_layout \end_inset \end_layout \begin_layout Plain Layout \end_layout \end_inset \end_layout \begin_layout Standard Based on these distributions, we selected a detection threshold of -1, which is approximately the leftmost edge of the trough between the signal and noise peaks. This represents the most liberal possible detection threshold that doesn't call substantial numbers of noise genes as detected. Among the full dataset, 13429 genes were detected at this threshold, and 22276 were not. When considering the GB libraries and non-GB libraries separately and re-comput ing normalization factors independently within each group, 14535 genes were detected in the GB libraries while only 12460 were detected in the non-GB libraries. Thus, GB allowed the detection of 2000 extra genes that were buried under the noise floor without GB. This pattern of at least 2000 additional genes detected with GB was also consistent across a wide range of possible detection thresholds, from -2 to 3 (see Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:Gene-detections" plural "false" caps "false" noprefix "false" \end_inset ). \end_layout \begin_layout Subsection Globin blocking does not add significant additional noise or decrease sample quality \end_layout \begin_layout Standard One potential worry is that the globin blocking protocol could perturb the levels of non-globin genes. There are two kinds of possible perturbations: systematic and random. The former is not a major concern for detection of differential expression, since a 2-fold change in every sample has no effect on the relative fold change between samples. In contrast, random perturbations would increase the noise and obscure the signal in the dataset, reducing the capacity to detect differential expression. \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/Globin Paper/figure4 - maplot-colored.pdf lyxscale 50 width 60col% groupId colwidth \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \begin_inset Argument 1 status collapsed \begin_layout Plain Layout MA plot showing effects of globin blocking on each gene's abundance. \end_layout \end_inset \begin_inset CommandInset label LatexCommand label name "fig:MA-plot" \end_inset \series bold MA plot showing effects of globin blocking on each gene's abundance. \series default All libraries were normalized together as described in Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:logcpm-dists" plural "false" caps "false" noprefix "false" \end_inset , and genes with an average logCPM below -1 were filtered out. Each remaining gene was tested for differential abundance with respect to globin blocking (GB) using \begin_inset Flex Code status open \begin_layout Plain Layout edgeR \end_layout \end_inset ’s quasi-likelihood F-test, fitting a negative binomial generalized linear model to table of read counts in each library. For each gene, \begin_inset Flex Code status open \begin_layout Plain Layout edgeR \end_layout \end_inset reported average abundance (logCPM), \begin_inset Formula $\log_{2}$ \end_inset fold change (logFC), p-value, and Benjamini-Hochberg adjusted false discovery rate (FDR). Each gene's logFC was plotted against its logCPM, colored by FDR. Red points are significant at ≤10% FDR, and blue are not significant at that threshold. The alpha and beta globin genes targeted for blocking are marked with large triangles, while all other genes are represented as small points. \end_layout \end_inset \end_layout \begin_layout Plain Layout \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Standardize on \begin_inset Quotes eld \end_inset log2 \begin_inset Quotes erd \end_inset notation \end_layout \end_inset \end_layout \begin_layout Standard The data do indeed show small systematic perturbations in gene levels (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:MA-plot" plural "false" caps "false" noprefix "false" \end_inset ). Other than the 3 designated alpha and beta globin genes, two other genes stand out as having especially large negative log fold changes: HBD and LOC1021365. HBD, delta globin, is most likely targeted by the blocking oligos due to high sequence homology with the other globin genes. LOC1021365 is the aforementioned ncRNA that is reverse-complementary to one of the alpha-like genes and that would be expected to be removed during the globin blocking step. All other genes appear in a cluster centered vertically at 0, and the vast majority of genes in this cluster show an absolute log2(FC) of 0.5 or less. Nevertheless, many of these small perturbations are still statistically significant, indicating that the globin blocking oligos likely cause very small but non-zero systematic perturbations in measured gene expression levels. \end_layout \begin_layout Standard \begin_inset Float figure wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Graphics filename graphics/Globin Paper/figure5 - corrplot.pdf lyxscale 50 width 70col% \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset Argument 1 status collapsed \begin_layout Plain Layout Comparison of inter-sample gene abundance correlations with and without globin blocking. \end_layout \end_inset \begin_inset CommandInset label LatexCommand label name "fig:gene-abundance-correlations" \end_inset Comparison of inter-sample gene abundance correlations with and without globin blocking (GB). \series default All libraries were normalized together as described in Figure 2, and genes with an average abundance (logCPM, log2 counts per million reads counted) less than -1 were filtered out. Each gene’s logCPM was computed in each library using the \begin_inset Flex Code status open \begin_layout Plain Layout edgeR \end_layout \end_inset cpm function. For each pair of biological samples, the Pearson correlation between those samples' GB libraries was plotted against the correlation between the same samples’ non-GB libraries. Each point represents an unique pair of samples. The solid gray line shows a quantile-quantile plot of distribution of GB correlations vs. that of non-GB correlations. The thin dashed line is the identity line, provided for reference. \end_layout \end_inset \end_layout \begin_layout Plain Layout \end_layout \end_inset \end_layout \begin_layout Standard To evaluate the possibility of globin blocking causing random perturbations and reducing sample quality, we computed the Pearson correlation between logCPM values for every pair of samples with and without GB and plotted them against each other (Figure \begin_inset CommandInset ref LatexCommand ref reference "fig:gene-abundance-correlations" plural "false" caps "false" noprefix "false" \end_inset ). The plot indicated that the GB libraries have higher sample-to-sample correlati ons than the non-GB libraries. Parametric and nonparametric tests for differences between the correlations with and without GB both confirmed that this difference was highly significant (2-sided paired t-test: t = 37.2, df = 665, P ≪ 2.2e-16; 2-sided Wilcoxon sign-rank test: V = 2195, P ≪ 2.2e-16). Performing the same tests on the Spearman correlations gave the same conclusion (t-test: t = 26.8, df = 665, P ≪ 2.2e-16; sign-rank test: V = 8781, P ≪ 2.2e-16). The \begin_inset Flex Code status open \begin_layout Plain Layout edgeR \end_layout \end_inset package was used to compute the overall biological coefficient of variation (BCV) for GB and non-GB libraries, and found that globin blocking resulted in a negligible increase in the BCV (0.417 with GB vs. 0.400 without). The near equality of the BCVs for both sets indicates that the higher correlati ons in the GB libraries are most likely a result of the increased yield of useful reads, which reduces the contribution of Poisson counting uncertainty to the overall variance of the logCPM values \begin_inset CommandInset citation LatexCommand cite key "McCarthy2012" literal "false" \end_inset . This improves the precision of expression measurements and more than offsets the negligible increase in BCV. \end_layout \begin_layout Subsection More differentially expressed genes are detected with globin blocking \end_layout \begin_layout Standard \begin_inset Float table wide false sideways false status collapsed \begin_layout Plain Layout \align center \begin_inset Tabular \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \series bold No Globin Blocking \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \series bold Up \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \series bold NS \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \series bold Down \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \series bold Globin-Blocking \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \series bold Up \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 231 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 515 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 2 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \series bold NS \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 160 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 11235 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 136 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \series bold Down \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 0 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 548 \end_layout \end_inset \begin_inset Text \begin_layout Plain Layout \family roman \series medium \shape up \size normal \emph off \bar no \strikeout off \xout off \uuline off \uwave off \noun off \color none 127 \end_layout \end_inset \end_inset \end_layout \begin_layout Plain Layout \begin_inset Caption Standard \begin_layout Plain Layout \series bold \begin_inset Argument 1 status open \begin_layout Plain Layout Comparison of significantly differentially expressed genes with and without globin blocking. \end_layout \end_inset \begin_inset CommandInset label LatexCommand label name "tab:Comparison-of-significant" \end_inset Comparison of significantly differentially expressed genes with and without globin blocking. \series default Up, Down: Genes significantly up/down-regulated in post-transplant samples relative to pre-transplant samples, with a false discovery rate of 10% or less. NS: Non-significant genes (false discovery rate greater than 10%). \end_layout \end_inset \end_layout \begin_layout Plain Layout \end_layout \end_inset \end_layout \begin_layout Standard To compare performance on differential gene expression tests, we took subsets of both the GB and non-GB libraries with exactly one pre-transplant and one post-transplant sample for each animal that had paired samples available for analysis (N=7 animals, N=14 samples in each subset). The same test for pre- vs. post-transplant differential gene expression was performed on the same 7 pairs of samples from GB libraries and non-GB libraries, in each case using an FDR of 10% as the threshold of significance. Out of 12954 genes that passed the detection threshold in both subsets, 358 were called significantly differentially expressed in the same direction in both sets; 1063 were differentially expressed in the GB set only; 296 were differentially expressed in the non-GB set only; 2 genes were called significantly up in the GB set but significantly down in the non-GB set; and the remaining 11235 were not called differentially expressed in either set. These data are summarized in Table \begin_inset CommandInset ref LatexCommand ref reference "tab:Comparison-of-significant" plural "false" caps "false" noprefix "false" \end_inset . The differences in BCV calculated by EdgeR for these subsets of samples were negligible (BCV = 0.302 for GB and 0.297 for non-GB). \end_layout \begin_layout Standard The key point is that the GB data results in substantially more differentially expressed calls than the non-GB data. Since there is no gold standard for this dataset, it is impossible to be certain whether this is due to under-calling of differential expression in the non-GB samples or over-calling in the GB samples. However, given that both datasets are derived from the same biological samples and have nearly equal BCVs, it is more likely that the larger number of DE calls in the GB samples are genuine detections that were enabled by the higher sequencing depth and measurement precision of the GB samples. Note that the same set of genes was considered in both subsets, so the larger number of differentially expressed gene calls in the GB data set reflects a greater sensitivity to detect significant differential gene expression and not simply the larger total number of detected genes in GB samples described earlier. \end_layout \begin_layout Section Discussion \end_layout \begin_layout Standard The original experience with whole blood gene expression profiling on DNA microarrays demonstrated that the high concentration of globin transcripts reduced the sensitivity to detect genes with relatively low expression levels, in effect, significantly reducing the sensitivity. To address this limitation, commercial protocols for globin reduction were developed based on strategies to block globin transcript amplification during labeling or physically removing globin transcripts by affinity bead methods \begin_inset CommandInset citation LatexCommand cite key "Winn2010" literal "false" \end_inset . More recently, using the latest generation of labeling protocols and arrays, it was determined that globin reduction was no longer necessary to obtain sufficient sensitivity to detect differential transcript expression \begin_inset CommandInset citation LatexCommand cite key "NuGEN2010" literal "false" \end_inset . However, we are not aware of any publications using these currently available protocols the with latest generation of microarrays that actually compare the detection sensitivity with and without globin reduction. However, in practice this has now been adopted generally primarily driven by concerns for cost control. The main objective of our work was to directly test the impact of globin gene transcripts and a new globin blocking protocol for application to the newest generation of differential gene expression profiling determined using next generation sequencing. \end_layout \begin_layout Standard The challenge of doing global gene expression profiling in cynomolgus monkeys is that the current available arrays were never designed to comprehensively cover this genome and have not been updated since the first assemblies of the cynomolgus genome were published. Therefore, we determined that the best strategy for peripheral blood profiling was to do deep RNA-seq and inform the workflow using the latest available genome assembly and annotation \begin_inset CommandInset citation LatexCommand cite key "Wilson2013" literal "false" \end_inset . However, it was not immediately clear whether globin reduction was necessary for RNA-seq or how much improvement in efficiency or sensitivity to detect differential gene expression would be achieved for the added cost and work. \end_layout \begin_layout Standard We only found one report that demonstrated that globin reduction significantly improved the effective read yields for sequencing of human peripheral blood cell RNA using a DeepSAGE protocol \begin_inset CommandInset citation LatexCommand cite key "Mastrokolias2012" literal "false" \end_inset . The approach to DeepSAGE involves two different restriction enzymes that purify and then tag small fragments of transcripts at specific locations and thus, significantly reduces the complexity of the transcriptome. Therefore, we could not determine how DeepSAGE results would translate to the common strategy in the field for assaying the entire transcript population by whole-transcriptome 3’-end RNA-seq. Furthermore, if globin reduction is necessary, we also needed a globin reduction method specific to cynomolgus globin sequences that would work an organism for which no kit is available off the shelf. \end_layout \begin_layout Standard As mentioned above, the addition of globin blocking oligos has a very small impact on measured expression levels of gene expression. However, this is a non-issue for the purposes of differential expression testing, since a systematic change in a gene in all samples does not affect relative expression levels between samples. However, we must acknowledge that simple comparisons of gene expression data obtained by GB and non-GB protocols are not possible without additional normalization. \end_layout \begin_layout Standard More importantly, globin blocking not only nearly doubles the yield of usable reads, it also increases inter-sample correlation and sensitivity to detect differential gene expression relative to the same set of samples profiled without blocking. In addition, globin blocking does not add a significant amount of random noise to the data. Globin blocking thus represents a cost-effective way to squeeze more data and statistical power out of the same blood samples and the same amount of sequencing. In conclusion, globin reduction greatly increases the yield of useful RNA-seq reads mapping to the rest of the genome, with minimal perturbations in the relative levels of non-globin genes. Based on these results, globin transcript reduction using sequence-specific, complementary blocking oligonucleotides is recommended for all deep RNA-seq of cynomolgus and other nonhuman primate blood samples. \end_layout \begin_layout Section Future Directions \end_layout \begin_layout Standard One drawback of the globin blocking method presented in this analysis is a poor yield of genic reads, only around 50%. In a separate experiment, the reagent mixture was modified so as to address this drawback, resulting in a method that produces an even better reduction in globin reads without reducing the overall fraction of genic reads. However, the data showing this improvement consists of only a few test samples, so the larger data set analyzed above was chosen in order to demonstra te the effectiveness of the method in reducing globin reads while preserving the biological signal. \end_layout \begin_layout Standard The motivation for developing a fast practical way to enrich for non-globin reads in cyno blood samples was to enable a large-scale RNA-seq experiment investigating the effects of mesenchymal stem cell infusion on blood gene expression in cynomologus transplant recipients in a time course after transplantation. With the globin blocking method in place, the way is now clear for this experiment to proceed. \end_layout \begin_layout Chapter Future Directions \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout If there are any chapter-independent future directions, put them here. Otherwise, delete this section. Check in the directions if this is OK. \end_layout \end_inset \end_layout \begin_layout Chapter Closing remarks \end_layout \begin_layout Standard \begin_inset ERT status collapsed \begin_layout Plain Layout % Use "References" as the title of the Bibliography \end_layout \begin_layout Plain Layout \backslash renewcommand{ \backslash bibname}{References} \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset CommandInset bibtex LatexCommand bibtex btprint "btPrintCited" bibfiles "code-refs,refs-PROCESSED" options "bibtotoc,unsrt" \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Check bib entry formatting & sort order \end_layout \end_inset \end_layout \begin_layout Standard \begin_inset Flex TODO Note (inline) status open \begin_layout Plain Layout Check in-text citation format. Probably don't just want [1], [2], etc. \end_layout \end_inset \end_layout \end_body \end_document