Fig. 2

Mouse full-stack state enrichments for external genomic annotations. A Fold enrichments of mouse full-stack states with external genome annotations (“Methods” section). Each row corresponds to a state and each column corresponds to one external genomic annotation: coding sequences, CpG islands, exons, gene bodies (exons and introns), transcription end sites (TES), transcription start sites (TSS), TSS and 2-kb surrounding regions, assembly gaps, pseudogenes, blacklisted regions, repeat elements, annotated Zfp genes, and PhastCons conserved elements (“Methods” section). The last row shows the percentage of the genome that each external genome annotation covers. The heatmap colors are column-normalized, i.e., within each column, the colors of the cells are such that the highest values are colored red and the lowest values are colored white. B Each row indicates the states’ average LECIF score, indicating evidence at the functional genomics level of human-mouse conservation based on epigenetic annotations [22] (“Methods” section). The list of states with top average LECIF scores and highest enrichments with PhastCons elements is in Additional file 1: Fig. S9 and Additional file 4. C Average weighted expression of genes that overlap each full-stack state in different groups of cell types (“Methods” section). Each column in the heatmap corresponds to a cell group indicated at the top. Each row corresponds to a state, as ordered in A. D, E Positional enrichments of full-stack states relative to annotated D transcription start sites (TSS) and E transcription end sites (TES). Positive coordinate values represent the number of bases downstream in the 5′ to 3′ direction of transcription, while negative values represent the number of bases upstream. Each line shows the positional enrichments in a state. Lines are colored corresponding to the state group as indicated in A. F Geometric mean and geometric standard deviation of enrichments of full-stack states CTCF elements across 28 cell types (“Methods” section). States are displayed vertically in the same order as A. The mOpenC6-7 state showed the strongest enrichment for CTCF elements in all observed cell types. The geometric mean and standard deviation are calculated such that for each state, fold enrichment values of 0 are replaced by the state’s minimum non-zero value. The fold enrichment values accompanying this plot are available in Additional file 4