The flexibility model can recapitulate the H2A.Z presence pattern bordering transcriptional start sites observed in vivo. A. H2A.Z occupancy calculated from data originating from Barski et al. Occupancy is calculated by counting how many times a nucleosome is found on each base pair. B. H2A occupancy, calculated as in panel A. C. Scores of the H2A.Z and H2A flexibility models over all human transcriptional start sites. Those regions are clearly a better fit for the H2A.Z model. D. Model and occupancy ratios. The occupancy ratio curve is calculated by dividing the occupancy counts of the H2A dataset by that of the H2A.Z dataset. The score ratio is calculated similarly by using the scores in panel C. The normalized curve is calculated by scaling the values of the score ratio curve between 0 and 1. The occupancy ratio curve was not normalized in any way.