Figure 2

Promoter-proximal PolII profiles of expressed transcripts can be modeled gene-by-gene. Quantification of PolII bimodal peaks in individual promoters. (a) An example of fitting a linear combination of two Gaussian profiles to the PolII ChIP-seq signal near the TSS of Ubxn2b. The parameters for this example are: inter-peak distance 113 bp, upstream peak width 48 bp, downstream peak width 18 bp, peak-height ratio 2.16, difference in occupancy + 28% for second peak, shift from annotated TSS 2 bp and quality of fit 0.99. (b) Quality of the model fit correlates well with the mRNA microarray signal. Ubxn2b is highlighted with a red dot (6,475 isolated genes). (c) Overhead view of normalized profiles of PolII at TSS after re-centering all promoters according to the TSS predicted by the model (1,475 expressed genes with quality score >0.85). (d) Distribution of the inter-peak distances and peak positions (same genes as (c)). (e) Distribution of the fitted width (standard deviation) of profile features (same genes as (c)). (f) Distribution of the difference in occupancy in upstream peak defined as (area under the upstream peak minus area under the downstream peak)/max(area under the upstream peak, area under the downstream peak) (same genes as (c)). a.u., arbitrary units.