2.5. Transcription Factor Enrichment Analysis (TFEA)

TFEA using the R package TFEA.ChIP [26] was used to identify transcription factors (TF) responsible fo the co-regulation of differentially expressed genes (DEGs) shared across RNA-seq datasets as well as for DEGs identified uniquely in a single dataset. TFEA.ChIP predicts transcription factor binding sites (TFBS) proximal to a given set of genes (DEGs) based on experimentally determined protein-DNA binding profiles from chromatin immunoprecipitation (ChIP) sequencing (ChIP-seq). A total of 1122 ChIP-seq datasets covering 333 different TFs generated by the ENCODE Consortium [27], as well as collected from the Gene Expression Omnibus (GEO), comprise the TFBS database used by TFEA.ChIP. The analysis of the association of TFBS consisted of comparing how many targets of a given TF are in two lists of genes of interest (DEGs versus non-DEGs). A contingency matrix was generated for each ChIP-seq in the TFBS database counting the number of genes it interacts with in the DEGs list and non-DEGs list, with a subsequent Fisher’s exact test applied to each contingency matrix to determine if the difference in the distribution is statistically significant. The results were subsequently summarized by TF. Gene Set Enrichment Analysis (GSEA) is then performed to test whether ChIP-seq datasets belonging to the same TF are significantly enriched or depleted as a group. Enrichment Score (ES) values reflect the degree to which a TF is over-represented at the top (up-regulated) or bottom (down-regulated) of a ranked list. The enriched TF ranking as well as the number of ChIP-seq datasets contributing to the TF enrichment were subsequently graphed with ggplot2 in R.