IntePareto: an R package for integrative analyses of RNA-Seq and ChIP-Seq data

Yingying Cao,Daniel Hoffmann,Simo Kitanovski

doi:10.1186/s12864-020-07205-6

Abstract

BackgroundRNA-Seq, the high-throughput sequencing (HT-Seq) of mRNAs, has become an essential tool for characterizing gene expression differences between different cell types and conditions. Gene expression is regulated by several mechanisms, including epigenetically by post-translational histone modifications which can be assessed by ChIP-Seq (Chromatin Immuno-Precipitation Sequencing). As more and more biological samples are analyzed by the combination of ChIP-Seq and RNA-Seq, the integrated analysis of the corresponding data sets becomes, theoretically, a unique option to study gene regulation. However, technically such analyses are still in their infancy.ResultsHere we introduce intePareto, a computational tool for the integrative analysis of RNA-Seq and ChIP-Seq data. With intePareto we match RNA-Seq and ChIP-Seq data at the level of genes, perform differential expression analysis between biological conditions, and prioritize genes with consistent changes in RNA-Seq and ChIP-Seq data using Pareto optimization.ConclusionintePareto facilitates comprehensive understanding of high dimensional transcriptomic and epigenomic data. Its superiority to a naive differential gene expression analysis with RNA-Seq and available integrative approach is demonstrated by analyzing a public dataset.

Highlights

RNA-Seq, the high-throughput sequencing (HT-Seq) of mRNAs, has become an essential tool for characterizing gene expression differences between different cell types and conditions
The consistent co-occurrence of histone modification patterns and up- or down-regulated gene expression can improve our understanding of the “histone code” [4]; or, the comparison of histone modification states with quantitative gene expression can lead to the discovery of new enhancer regions [5]; or, expression and simultaneous occurrence of different modifications at a gene can reveal gene regulation dynamics along a developmental trajectory [6]
We demonstrate that integration of RNA-Seq data and ChIP-Seq data by Pareto Optimization outperforms a clustering method based on Bayesian inference of a hierarchical model [11], and the analysis of RNA-Seq alone

Summary

Results

Evaluation of intePareto using publicly available data RNA-Seq and ChIP-Seq data We evaluate intePareto based on publicly available RNASeq and ChIP-Seq data from a study of Tet methylcytosine dioxygenase 2 (Tet2) knockout mouse embryonic stem cells (mESCs) that are compared to wild type mESCs [32]. The GO terms of interest were those confirmed in previous research such as “neurogenesis” [32, 40], “cardiac chamber development” [41, 42], “mammary gland formation” [43, 45], and “limb morphogenesis” [46] Both our integrative approach and the model-based approach found that the genes in the top-ranked genes were enriched in “neurogenesis” (Fig. 4a) and “limb morphogenesis” (Fig. 4d). IntePareto found that the top-ranked genes are more enriched in “cardiac chamber development” (Fig. 4b) and “mammary gland formation” (Fig. 4c) as they should be These functions were not identified by RNA-Seq analysis alone or the model-based approach. An alternative to GO enrichment, that yields complementary information, is pathway enrichment

Background

Discussion and conclusions