Fig. 3

Identification of shared and type-dependent EBNA2 human cofactors. A Unbiased computational prediction of EBNA2 human cofactors. Human transcription factor (hTF) motif enrichment analysis was performed within EBNA2 type 1 specific, type 2 specific, and shared peaks. Each dot represents the normalized significance of one hTF motif. Type 1 specific (y-axis, left) or type 2 specific (y-axis, right) normalized motif significance is compared to shared peaks (x-axis in both panels). Black diamonds indicate exemplar hTF motifs for the four hTF classes that are depicted in (B). The black line indicates equivalent significance between the compared peak sets. Motifs are colored by class. B Frequency of occurrence of exemplary motifs in EBNA2 shared and type-specific peak sets. For the four exemplary motifs, the percent of peaks containing predicted binding sites for the motif is shown. Each bar represents percent foreground (i.e., the percent of actual peak DNA sequences containing a match to the motif). The horizontal black line within each bar depicts percent background (i.e., the percent of randomly selected genome sequences, matching GC content). Asterisks indicate significant motif enrichment (P < 0.05), as calculated by HOMER. C Experimental validation of predicted EBNA2 co-occupancy with hTFs. Co-occupancy was assessed by hTF and EBNA2 ChIP-seq peak overlap. The hTFs BATF and JUNB were chosen as representative AP-1 family members (see Results). For each of the five hTFs, a union peak set was created by combining peaks across all cell lines. For each bar, the percent of each hTF union peak set overlapping each EBNA2 ChIP-seq peak category is shown. Datasets with significant overlap between EBNA2 peak sets and the union peak set of hTFs (as calculated by RELI) are indicated with asterisks (P < 0.05). Note the consistency between the motif-based predictions (B) and ChIP-seq experimental validation results (C)