Figure 12

TF bindings sites are clustered (or “stacked”) within promoters. Two theoretically possible alternative Tfbs distributions in proximal promoters are illustrated in panels a and c. Panel a illustrates a situation where Tfbs are located in a “stacked” arrangement close to the TSS. This arrangement results in high percentage overlap in pairwise Tfbs comparisons. In contrast, panel c illustrates a situation where Tfbs are randomly distributed in the proximal promoter. To test for these two trends on the whole genome scale, we calculated percentage overlap in all pairwise Tfbs comparisons in each promoter (1 kb) for 25,930 RefSeq transcripts (panel b ). The observed overlap (corresponding to the “stacked” arrangement of Tfbs illustrated in panel a) was contrasted against a randomized dataset where Tfbs were assigned random positions within the same proximal promoter (corresponding to the random arrangement of Tfbs illustrated in panel c and resulting in a lower percentage overlap in pairwise Tfbs comparisons). As an example, we demonstrate the SRSF2/MFSD11 locus in panel d, which has the highest number of Tfbs (i.e. 71) in our dataset, arranged in the overlapping or “stacked” mode.