CexoR: an R/Bioconductor package to uncover high-resolution protein-DNA interactions in ChIP-exo replicates

Abstract

For its unprecedented level of spatial resolution, chromatin immunoprecipitation combined with λ exonuclease digestion followed by high-throughput sequencing (ChIP-exo) has the potential to replace ChIP-seq as the standard approach for genome-wide mapping of protein-DNA interactions. In this assay, the midpoint between the strand-specific paired peaks, formed in the forward and reverse strands, is typically delimited by the exonuclease stop-sites, within which the protein-binding events are located. Although numerous algorithms have been developed for peak-calling in ChIP-seq data, none of them is fully adjusted for the analysis of ChIP-exo. This is because those statistical models do not make use of ChIP-exo’s strand-specificity for the identification of protein-DNA binding sites. Here, we present the CexoR algorithm, which aims to ease the analysis of replicated ChIP-exo data in BAM alignment format. The detection algorithm relies on the Skellam distribution (cross-correlation of two Poisson distributions) to calculate probabilities of consecutive punctate-sources of read-enrichment located nearby at Watson-and-Crick strands. ChIP-exo peak-pairs are identified and ranked by their irreproducible discovery rate estimated across biological replicates, and finally reported in BED format files. CexoR can potentially be applied to other ChIP-exo-based protocols, such as ChIP-nexus.

Availability and implementation:

CexoR has been implemented in R, and is freely available at http://bioconductor.org.