A reference methylome database and analysis pipeline to facilitate integrative and comparative epigenomics.
Song Q, Decato B, Hong EE, Zhou M, Fang F, Qu J, Garvin T, Kessler M, Zhou J, Smith AD. · 2013
Read A reference methylome database and analysis pipeline to facilitate integrative and comparative epigenomics. by Song Q, Decato B, Hong EE, Zhou M, Fang F, Qu J, Garvin T, Kessler M, Zhou J, Smith AD. (2013) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc3855694.txt.
Introduction
Epigenome tools are the offering of Qiang Song and co-authors in PLoS One in 2013. DNA methylation is a chemical mark that helps control gene use. New sequencing reads those marks base by base across whole genomes, and public data sets now pile up fast.
The team built two linked resources. MethPipe is a software path that turns raw reads into mark maps and pulls out key features at low and high levels. MethBase is a store of public mark maps, each with notes, ready for side-by-side checks. Together they let a lab compare its own maps with the public stock in a few steps. The authors aim to speed shared, cross-study work in a fast-growing field.
Catalog record
| CREATOR | Song Q, Decato B, Hong EE, Zhou M, Fang F, Qu J, Garvin T, Kessler M, Zhou J, Smith AD. |
|---|---|
| YEAR | 2013 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 41 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A Reference Methylome Database and Analysis Pipeline to Facilitate Integrative and Comparative Epigenomics By Qiang Song, Benjamin Decato, Elizabeth E Hong, Meng Zhou, Fang Fang, Jianghan Qu, Tyler Garvin, Michael Kessler, Jun Zhou, Andrew D Smith, Osman El-Maarri Published in: PLoS ONE DOI: 10.1371/journal.pone.0081148 ABSTRACT Abstract DNA methylation is implicated in a surprising diversity of regulatory, evolutionary processes and diseases in eukaryotes. The introduction of whole-genome bisulfite sequencing has enabled the study of DNA methylation at a single-base resolution, revealing many new aspects of DNA methylation and highlighting the usefulness of methylome data in understanding a variety of genomic phenomena. As the number of publicly available whole-genome bisulfite sequencing studies reaches into the hundreds, reliable and convenient tools for comparing and analyzing methylomes become increasingly important. We present MethPipe, a pipeline for both low and high-level methylome analysis, and MethBase, an accompanying database of annotated methylomes from the public domain. Together these resources enable researchers to extract interesting features from methylomes and compare them with those identified in public methylomes in our database. INTRODUCTION In eukaryotes, DNA methylation refers to the addition of a methyl group to the 5′ location of cytosines and is closely linked to transcriptional regulation [1]. Lack of methylation in genomic regions usually indicates the accessibility of the underlying DNA sequences to transcription factor binding [2], although recent results have suggested a more complex relationship [3]. Investigating DNA methylation patterns across the entire genomes is a key step to understanding its various roles. There are several techniques to profile DNA methylation, the most comprehensive being whole-genome bisulfite sequencing (WGBS; [4]), which has been used to profile methylation levels across entire genomes in a variety of organisms and cell types [5]–[7]. Alternatively, reduced representation bisulfite sequencing (RRBS) preferentially profiles CpG-rich regions and requires less sequencing [8]. These profiling methods provide single-based resolution for the full epigenome unattainable with ChIP-seq and DNase-seq. For example, recent studies have identified subtle boundary changes in hypo-methylated regions around promoters during cell differentiation – discoveries that would not have been possible without single-base resolution profiling [9], [10]. These experiments provide information for numerous regions of dynamic methylation, believed to related with promoters, enhancers, insulators and more broadly regions dense in transcription factor binding sites [11]. Comparative analysis of the methylomes of different cell types and conditions reveals functional epigenetic domains with implications in cell differentiation and disease onset [10], [12], [13]. The analyses and interpretation of bisulfite sequencing data are usually performed in a multi-stage, multi-resolution manner. Extensive efforts have been devoted to mapping reads and estimating methylation levels at individual cytosine sites [14], [15]. Meanwhile, growing interest is shifting towards biologically meaningful higher-level methylation features, such as hypo-methylated regions (HMRs; [3], [9]), large-scale partially methylated domains (PMDs; [7], [13]), and allele-specific methylated regions (AMRs; [16], [17]). Existing methods to identify these are usually project specific, with few general tools available for down-stream analysis tasks. As the number of publicly available methylomes rises, large-scale comparative analysis of methylation patterns across multiple samples calls for well-curated reference methylome databases. Some databases exist to address the need for accessing bisulfite sequencing datasets. The NCBI Sequence Read Archive (SRA) contains raw sequences of most BS-seq data [18]. The NGSmethDB database [19] and the NCBI Epigenomics Resources [20] provide pre-computed methylation level results at individual cytosines, saving both time and computational resources to download and analyze those raw reads,. However there are not sufficient effort to address the issue of higher-level methylation features. This creates a computational barrier for researchers who seek to use methylation-based features in large projects. Additionally, databases like SRA and NCBI Epigenetic Resources are designed to accommodate datasets generated from a variety of techniques, including ChIP-seq, RNA-seq and WGBS. As such, their annotation lacks metadata (e.g. bisulfite conversion rates) that are specific to bisulfite sequencing. MethBase addresses the need for a well-annotated database of methylome features and summary statistics (http://smithlab.usc.edu/methbase/). We collected bisulfite sequencing datasets from SRA and other primary sources and analyzed each with the MethPipe pipeline. From there, we integrated detailed meta information into the UCSC Genome Browser. This provides a convenient avenue through which researchers may select methylomes of interest, visualize them in the UCSC Genome Browser, examine more detailed metadata, and download all available information for additional analysis. MethPipe is a standalone pipeline and freely available open source software, and provides for easy and consistent comparison between private data and public methylomes. Here we explain several aspects of these resources. We also explore a set of general observations that emerge naturally from these resources, demonstrating their power and usefulness in comparative analysis, data integration and technical benchmarking of newly produced methylome datasets. RESULTS Types of Data in the Database Individual records in MethBase, referred to here simply as methylomes, correspond to biological samples or replicates, as defined in the original publication of the raw data. When datasets are defined as different technical replicates, for example different libraries made from the same sample, they are pooled in MethBase as one entry. For each methylome, the database provides a variety of data, such as methylation levels of individual cytosines, hypo-methylated regions, partially methylated domains, allele-specific methylated regions, and detailed metadata and summary statistics. We describe these types of data below, and reserve more technical discussion of the pipeline used to generate these data in the Methods section. Methylation levels Although individual methyl groups can either be present or absent on a cytosine, current WGBS data is not from single cells, and we therefore refer to a methylation “level” for each cytosine, interpreted as the fraction of molecules in the underlying cell population that have the methyl mark on the corresponding cytosine. Since bisulfite sequencing gives a readout for either the presence or absence of a methyl group in each read, we get an unbiased estimate for the methylation level from the ratio of methylated reads to all reads covering a given site. The estimated cytosine methylation level, along with the number of sequenced reads supporting that estimate, is computed for each methylome, along with confidence intervals as described by Hodges et al. [21]. Hypo-methylated regions Excluding the brief developmental stages of somatic and germ cell reprogramming [22], mammalian cells typically have high levels of methylation throughout the genome. The more interesting features of mammalian methylomes tend to be those regions lacking methylation (e.g. Figure 1A). These hypo-methylated regions (HMRs) have been associated with CpG islands, promoter regions, and more generally enhancers and insulators [11]. We developed a highly accurate method for identifying HMRs using a stochastic segmentation model that accounts for both changes in methylation levels and variance in read coverage along the genome [9]. An individual HMR could represent an interval of the genome that has been protected from methylation, or has been opened up by the activity of specific DNA binding proteins [3]. Regardless of the causes, these intervals are strongly associated with regulation of gene expression, and fluctuations in their boundaries between methylomes likely indicate context-specific regulatory sites [10]. In healthy somatic cell methylomes, the organization of methylation features exhibit strikingly precise boundaries. The precision with which HMRs can be identified is what separates this kind of epigenetic information from others, such as H3K4me marks or DNase