Cistrome-GO: a web server for functional enrichment analysis of transcription factor ChIP-seq peaks.

Shaojuan Li,Clifford A Meyer,Changxin Wan,Rongbin Zheng,Xin Dong,Jingyu Fan,X Shirley Liu

doi:10.1093/nar/gkz332

Shaojuan Li, Clifford A Meyer + Show 5 more

Open Access

https://doi.org/10.1093/nar/gkz332

Copy DOI

Abstract

Characterizing the ontologies of genes directly regulated by a transcription factor (TF), can help to elucidate the TF’s biological role. Previously, we developed a widely used method, BETA, to integrate TF ChIP-seq peaks with differential gene expression (DGE) data to infer direct target genes. Here, we provide Cistrome-GO, a website implementation of this method with enhanced features to conduct ontology analyses of gene regulation by TFs in human and mouse. Cistrome-GO has two working modes: solo mode for ChIP-seq peak analysis; and ensemble mode, which integrates ChIP-seq peaks with DGE data. Cistrome-GO is freely available at http://go.cistrome.org/.

Highlights

Insight into the biological roles of transcription factors (TFs) can be acquired through analyses of the ontologies of the genes that they regulate
MYOD1 is known to function as a master regulator in muscle cell differentiation, and most of its annotated Biological Process (BP) GO terms are related to muscle development
Despite the popularity of ChIP-seq technology and the importance of functional enrichment analysis, few tools are available for functional enrichment analysis based on ChIPseq peaks

Summary

Introduction

Insight into the biological roles of transcription factors (TFs) can be acquired through analyses of the ontologies of the genes that they regulate. Despite the popularity of ChIP-seq technology and the importance of functional enrichment analysis few web servers are available for functional enrichment analysis based on ChIP-seq peaks (Table 1) Both ChIPEnrich [2] and Enrichr [3] use a binary value to measure each gene’s potential to be regulated by a TF. For ChIP-seq data with many peaks (e.g. 20 000), tools based on binary value assignments could fail to identify the relevant enriched GO terms, due to too many genes being assigned as targets Their functional enrichment analysis approaches rely on given sharp yet arbitrary cutoffs for target assignment (e.g. within 10 kb to a TSS). None of the published web servers facilitate the integration of differential gene expression (DGE) data, such that derived from perturbation in the TF’s activity This integration can improve the accuracy of the target gene prediction and thereby improve the GO association analysis

Methods

Results

Conclusion