Figure 4. The intrinsic nucleosome organization over transcripts and transcription factor binding sites.

a, For the in vitro and in vivo nucleosome occupancy maps, and for our model, shown is the normalized nucleosome occupancy per base pair around the transcription start site, averaged across all yeast genes. The long-range ordering of nucleosome occupancy which is present in the in vivo maps but not in the in vitro map may be partly explained by the lower nucleosome concentration in which the in vitro experiment was carried out (see Methods), because higher nucleosome concentration in vivo is predicted to cause long-range ordering of nucleosome arrays²¹. b, Same as a, but around translation end sites of genes (translation end was chosen because transcription end sites are poorly annotated). The depletion around gene ends may be due to the presence of termination signals²⁷, which disfavour nucleosome formation in vitro (Supplementary Fig. 8). The fact that these signals tend to occur in a specific orientation with respect to the direction of transcription²⁷ is consistent with a function in transcript processing, but does not exclude the possibility that one or more of these motifs functions primarily to disfavour nucleosomes. c, Comparison of the nucleosome occupancy over transcription factor binding sites between the in vitro and the YPD in vivo maps. For each transcription factor with at least 50 functional binding sites²², we computed, separately for the in vivo and in vitro maps, the average normalized nucleosome occupancy over its binding sites. Shown is a comparison of these nucleosome occupancies per factor, between the in vivo and in vitro maps, along with the Pearson correlation between them. For six factors taken from different regions of the plot, we also show the average normalized nucleosome occupancy around those factors’ binding sites, for both the in vitro and the in vivo maps.