Integrating Peak Colocalization and Motif Enrichment Analysis for the Discovery of Genome-Wide Regulatory Modules and Transcription Factor Recruitment Rules

Abstract

Chromatin immunoprecipitation followed by next-generation sequencing (ChIP-Seq) has opened new avenues of research in the genome-wide characterization of regulatory DNAprotein interactions at the genetic and epigenetic level. As a consequence, it has become the de facto standard for studies on the regulation of transcription, and literally thousands of data sets for transcription factors and cofactors in different conditions and species are now available to the scientific community. However, while pipelines and best practices have been established for the analysis of a single experiment, there is still no consensus on the best way to perform an integrated analysis of multiple datasets in the same condition, in order to identify the most relevant and widespread regulatory modules composed by different transcription factors and cofactors. We present here a computational pipeline for this task, that integrates peak summit colocalization, a novel statistical framework for the evaluation of its significance, and motif enrichment analysis. We show examples of its application to ENCODE data, that led to the identification of relevant regulatory modules composed of different factors, as well as the organization on DNA of the binding motifs responsible for their recruitment.

On the NF-Y regulome as in ENCODE (2019)

“…Having gathered data with these stringent criteria, we initially performed two types of experiments: (i) evaluation of the enrichment of the CCAAT matrix in the peaks of ChIP-seq experiments of all TFs and cofactors, using the PscanChIP software. This method classifies motif enrichment as “Global” or “Local” and signals a positional bias of motifs–CCAAT in this case–with respect to peaks summits [ 28 ]. (ii) Computation of overlaps between NF-Y peaks and those of all other factors, irrespective of CCAAT enrichment with PscanChIP: this validates the enrichment of CCAAT, as it formally proves that the identified CCAAT are indeed NF-Y-bound, and it also identifies co-association with additional factors.…”

“…By so doing, we identified 19 TFs with a positional bias between motifs, as shown in Fig 2 with representative matrices, and, with all data, in S1 Fig . Some of these–Sp1, USF1, USF2, RUNX3, C/EBPB, MAFK, ZNF143– have been discussed before [ 12 , 17 – 19 , 21 , 28 ]. FOS, previously found on double CCAAT boxes, is now detected, as in a previous study [ 21 ], also with its consensus “TRE” matrix, 10/12 bp upstream of CCAAT.…”

“…Since in some cases different experiments were available for the same TF, we filtered this initial dataset as follows. Duplicate experiments for the same TF in the same cell line were processed as previously described [ 28 ]: first of all, a total of 277 duplicate experiments performed with antibodies directed against a Tagged protein were removed. We further discarded all experiments (minima) with less than 10000 peaks or less than half of the peaks of the other replicates for the same factor.…”

“…The computation and statistical evaluation of peak overlap was performed as recently described [ 28 ]. Briefly, the overlap between two ChIP-seq peak sets was computed by counting the number of summits of the first TF falling within 300 bp regions centered on the peak summits of the second.…”

“…That is, this value was estimated as the number of 150 bp-width DNaseI hypersensitive regions in the cell lines employed. With small differences, this value had an average of 250000, which is the value we employed in our calculations [ 12 , 28 ]. Statistical significance was evaluated according to a Poisson distribution, p-values were Bonferroni corrected, and the -Log 10 of the p-values was used for producing the clustered heatmaps.…”