Detecting differential peaks in ChIP-seq signals with ODIN

Abstract

We propose an One-stage DIffereNtial peak caller (ODIN); an Hidden Markov Model-based approach to detect and analyze differential peaks (DPs) in pairs of ChIP-seq data. ODIN performs genomic signal processing, peak calling and p-value calculation in an integrated framework. We also propose an evaluation methodology to compare ODIN with competing methods. The evaluation method is based on the association of DPs with expression changes in the same cellular conditions. Our empirical study based on several ChIP-seq experiments from transcription factors, histone modifications and simulated data shows that ODIN outperforms considered competing methods in most scenarios.

GR and ER Coactivation Alters the Expression of Differentiation Genes and Associates with Improved ER+ Breast Cancer Outcome

“…Bioinformatics analysis of GR ChIP-seq revealed GR binding regions (GBRs), indicated by ChIP-seq peaks. To reduce the occurrence of false positives, GR-associated chromatin sequences were identified using concordant peaks from two algorithms, ODIN (16) and MACS (17). In the absence of ligand, only 726 genome-wide GR peaks were observed (Supplementary Figure 1A).…”

“…The sequence alignment and identification of peaks is described briefly. Sequence quality was assessed and aligned to the human genome (version hg19), and peaks were detected using two algorithms: rgt-ODIN v0.3.2 (16) and MACS2 v2.1.0.20140616 (17). The final lists of peaks contained concordant peaks between the two algorithms, with a false discovery rate of 0.05 (MACS2).…”

Differential peak calling of ChIP-seq signals with replicates with THOR

“…First, their DPs are restricted to their initial candidate regions. Therefore, they fail to detect subtle changes within these candidate regions (25,30). This is particularly crucial for ChIP-seq data of histone modifications, where differential peaks occur in small regions within larger genomic regions with high ChIP-Seq signal.…”

“…One-stage DPC methods are based on segmentation methods, such as Hidden Markov models (HMMs) (20,25,28) or sliding window based approaches (21,22,24,26,27). These methods solve most of the issues not addressed by two-stage DPC methods.…”

“…While large windows will fail to detect small changes in peaks on histone modifications, small windows can generate overly fragmented peaks. Methods based on HMMs, which intrinsically segment the signals on windows with varying size, are capable of detecting subtle changes in DNA–protein interactions (20,25,28), i.e. partial gains/losses of histone modifications within large chromatin domains.…”