Fig. 3

Relationship of annotations to known genomic elements. a Enrichment of each label over an idealized gene. We calculated enrichment as the base-2 logarithm of the observed frequency of a label at a particular position along an annotation divided by the expected frequency of the label from its prevalence in the genome overall. Enriched positions are shown in red, and depleted positions are shown in blue. b Relationship of labels to selected input data sets. Color corresponds to the mean signal value of a given assay type at positions annotated with a given label, aggregated over cell types where the given assay type is available. Values are normalized such that the maximum and minimum in each column are 1 and 0, respectively. Columns are ordered by hierarchical clustering for readability. c Fraction of the genome covered by each label. d Distribution of conservation-associated activity scores for states with interpretation term. Note that the units of the histogram are states, not base pairs