Hygeia: A new framework for data-driven discovery of DNA methylation patterns
Track: Tiny Paper Track
Keywords: epigenetics, DNA methylation, Bayesian modelling, change-point model
TL;DR: We introduce flexible Bayesian change-point models for DNA methylation and associated inference algorithms that yield a detailed probabilistic description of methylation signatures.
Abstract: Epigenetic variability is an essential modulator of phenotypic plasticity. To better understand complex epigenetic signals, we introduce Hygeia – a new framework for discovering DNA methylation patterns in whole-genome bisulfite sequencing (WGBS) data. Hygeia utilises a Bayesian statistical model, designed to match empirically observed methylation patterns. The model selects a regime for the methylation propensity at each cytosine-guanine dinucleotide (CpG) site, with regime changes permitted at any position. Thus, conventional means-based methods are replaced by probability-based MEthylaTion changE point Regimes (METEORs). Hygeia fits the model to WGBS data to produce METEOR annotation at the CpG level. We applied Hygeia to WGBS EpiATLAS data (N=445) from the International Human Epigenome Consortium (IHEC) to enrich the EpiATLAS resource with METEOR annotation. Hygeia is packaged as a Nextflow pipeline available on GitHub.
Submission Number: 81
Loading