Poly-Enrich: Count-based Methods for Gene Set Enrichment Testing with Genomic Regions

Abstract

12Gene set enrichment (GSE) testing enhances the biological interpretation of ChIP-13 seq data and other large sets of genomic regions. Our group has previously 14 introduced two GSE methods for genomic regions: ChIP-Enrich for narrow regions 15 and Broad-Enrich for broad genomic regions, such as histone modifications. Here, 16we introduce new methods and extensions that more appropriately analyze sets of 17 genomic regions with vastly different properties. First, we introduce Poly-Enrich, 18Although every cell in our body contains the same DNA, our cells perform vastly 42 different functions due to differences in how our genes are regulated. Certain 43 regions of the genome are bound by DNA binding proteins (transcription factors), 44 which regulate the expression of nearby genes. After an experiment to identify a 45 large set of these regions, we can then model the association of these regions with 46 various cellular pathways and biological processes. This analysis helps understand 47 the overall biological effect that the binding events have on the cells. For example, if 48 genes relating to apoptosis tend to have the transcription factor, Bcl-2, bind more 49 often nearby, then Bcl-2 is likely to have a vital role in regulating apoptosis. The 50 specifics of how to best perform this analysis is still being researched and depends 51 on properties of the set of genomic regions. Here, we introduce a new, more flexible 52 method that counts the number of occurrences per gene and models that in a 53 sophisticated statistical test, and compare it to a previous method. We show that the 54 optimal method depends on multiple factors, and the new method, Poly-Enrich, 55 allows interesting findings in scenarios where the previous method failed. 56 57

Sex-Specific Programming of Cardiac DNA Methylation by Developmental Phthalate Exposure

“…PolyEnrich 46 was used to assess biological pathways enriched among the DMCs. Analyses were stratified by sex and direction of differential methylation.…”

Testing Proximity of Genomic Regions to Transcription Start Sites and Enhancers Complements Gene Set Enrichment Testing

“…We tested all 90 ENCODE datasets using the polyenrich method in the chipenrich Bioconductor package, using the "nearest_tss" gene locus definition and GO biological processes for the gene sets. Poly-Enrich performs GSE on sets of peaks by testing if the number of peaks regulating a gene set is greater or less than that not in the gene set, taking into account the number of peaks assigned to each gene (Lee et al, 2018). The statistical model uses a negative binomial glm with an adjustment for gene locus length.…”

“…In parallel to this, a standard GSE test is performed using the same gene sets. In this article, we applied the polyenrich method for the GSE test (Lee et al, 2018), but others may be used. Result files contain the proximity results with test direction (enriched/depleted from GSE, and closer/farther from ProxReg), p-values and FDR values.…”

“…Most GSE methods were developed for gene expression data, do not adjust for the varying lengths of genes or regulatory space between them, and thus are not generally appropriate for GSE testing with large sets of peaks. However, several GSE methods have been developed to specifically test sets of peaks, including GREAT (McLean et al, 2010), ChIP-Enrich (Welch et al, 2014), Broad-Enrich (Cavalcante et al, 2014), and Poly-Enrich (Lee et al, 2018). Among these, Poly-Enrich is the only method that counts genomic regions (which we will refer to as peaks for simplicity) for each gene, adjusts for the varying lengths of genes and regulatory space between them, and provides a flexible approach with the ability to assign weights to peaks.…”

“…Therefore, it is of great interest to know the patterns of TF binding with respect to promoters and enhancers of the target genes and pathways. Although GREAT (McLean et al, 2010), ChIP-Enrich (Welch et al, 2014), and Poly-Enrich (Lee et al, 2018) incorporate distal binding events in their GSE testing, no method has been established for answering the question of whether a TF is binding closer to TSSs, near enhancers, both, or neither for a specific gene set.…”