Figure 3

Characterization of common STAT binding sites. (A) Sequence conservation of shared STAT binding regions was calculated and averaged using PhastCons score. The x-axis indicates the number of overlaps by any of the STAT binding sites. Two random sets were generated (same number of regions – high / intermediate sets) for comparison. Pearson’s correlation coefficient r = 0.95 (high), r = 0.69 (intermediate), r = 0.04 (random1) and r = 0.53 (random2). (B) Distributions of STAT-associated CRMs and all STAT binding sites were estimated. The ‘others’ category includes 5^′- and 3^′-UTRs. (C) Statistically significant pathways associated with the genes near CRCCs were inferred using the GREAT tool with the following default settings; Basal plus extension – proximal: 5 kb upstream and 1kb downstream / distal: up to 1000 kb [28]. P-value is the Bonferroni corrected binomial P-value. (D) STATs 1, 3, 4, 5 and 6 can bind to the same sites in similar or different cell contexts. Six loci encoding Stat1, Socs2, Socs3, Cish, Ifnar2 and Irf9 are shown. Red vertical bars overlapping peaks indicate the binding regions of STATs 1, 3, 4, 5 and 6 shared between several cell types, while blue vertical bars represent the sites with unique context-dependent STAT binding. The bottom panel depicts positions of peak summits and sequence conservation as well as GAS motifs (TTCnnnGAA, perfect match) at the STAT binding regions. The total number of tags in ChIPed samples was normalized to the corresponding input using the wignorm program [24] and the y-axis scale was adjusted according to this normalized value using auto-scale function of the UCSC genome browser. Red asterisks denote peak centers.