Figure 2

Sequence-specific models of nucleosome occupancy. A) Position-independent models. N = 2 position-independent (PI) models fit on the datasets indicated were used to predict nucleosome occupancy profiles on the C. elegans genome. Pearson correlations between predicted occupancy profiles are plotted. For each comparison, a single correlation coefficient is calculated across all chromosomes. B) Bioinformatics versus position-independent models. Same as (A), but comparing predictions made by bioinformatics models (Bio) to those made by position-independent models. C) Spatially resolved versus position-independent models. Same as (A), but comparing predictions made by spatially resolved models (SR) to those made by N = 2 position-independent models. D) Two-dimensional histogram of nucleosome occupancy predicted by a position-independent in vivo model. Chromosome III was divided into one thousand segments of equal length. The relative position of each segment is shown on the x-axis. The nucleosome occupancy, as predicted by the N = 2 position-independent model fit on Gu & Fire in vivo data [38], is shown on the y-axis. The predicted occupancy was binned into 100 equal intervals in the [0,1] range. For each segment, a color-coded histogram shows the number of bases with a given occupancy. The black line shows mean predicted occupancy in each segment. The range of the y-axis excludes the top 1.0% of bases with the highest occupancy. E) Two-dimensional histogram of nucleosome occupancy predicted by a position-independent in vitro model. Same as (D), but with occupancy predicted by the N = 2 position-independent model fit on in vitro data from the Hinc II assay.