PEGS: An efficient tool for gene set enrichment within defined sets of genomic intervals

Peter Briggs,Andrew D Sharrocks,Mudassar Iqbal,Shen-Hsi Yang,A Louise Hunter

doi:10.12688/f1000research.53926.1

Abstract

Many biological studies of transcriptional control mechanisms produce lists of genes and non-coding genomic intervals from corresponding gene expression and epigenomic assays. In higher organisms, such as eukaryotes, genes may be regulated by distal elements, with these elements lying 10s–100s of kilobases away from a gene transcription start site. To gain insight into these distal regulatory mechanisms, it is important to determine comparative enrichment of genes of interest in relation to genomic regions of interest, and to be able to do so at a range of distances. Existing bioinformatics tools can annotate genomic regions to nearest known genes, or look for transcription factor binding sites in relation to gene transcription start sites. Here, we present PEGS (Peak set Enrichment in Gene Sets). This tool efficiently provides an exploratory analysis by calculating enrichment of multiple gene sets, associated with multiple non-coding elements (peak sets), at multiple genomic distances, and within topologically associated domains. We apply PEGS to gene sets derived from gene expression studies, and genomic intervals from corresponding ChIP-seq and ATAC-seq experiments to derive biologically meaningful results. We also demonstrate an extended application to tissue-specific gene sets and publicly available GWAS data, to find enrichment of sleep trait associated SNPs in relation to tissue-specific gene expression profiles.

Highlights

Gene expression control in higher organisms is achieved through a complex hierarchical process involving opening of chromatin, histone modifications, and binding of transcription factors (TFs)
We present a new tool – PEGS (Peak set Enrichment in Gene Sets)1 – which calculates mutual enrichment of multiple gene sets associated with multiple peak sets, simultaneously and efficiently
Gene clusters are provided as text files with one gene symbol on each line; genomic region coordinates are provided in standard BED format

Summary

Introduction

Gene expression control in higher organisms is achieved through a complex hierarchical process involving opening of chromatin, histone modifications, and binding of transcription factors (TFs). Depending on the design of the experiment, these analyses produce differentially expressed gene sets or clusters for further analysis. These studies are often complemented by assays which map, on a genome-wide scale, TF binding sites (ChIP-seq) or regions of chromatin accessibility (DNase-seq, ATAC-seq). Analyses of these data produce a collection of genomic intervals (peak sets). An important computational task is to integrate these data to produce meaningful results; i.e. to relate gene sets to peak sets to aid functional interpretation. Bearing in mind distal regulation, an important consideration here is to be able to calculate gene set enrichment at multiple genomic distances from peak sets, and to be able to do this efficiently within the same analysis

Methods

Results

Conclusion