Extended Data Fig. 8: Chromatin state developmental dynamics.

a, Enrichment of accessible chromatin within regions segmented into different chromatin states. Left, values for a set of ChromHMM annotations made using ChIP–seq data pooled from both biological replicates. Right, values for a more conservative set of ChromHMM annotations including only those regions annotated in the same state independently in both biological replicates. n = 66 tissue-stages per box. b, Hierarchical relationships among strong enhancers (state no. 5) in different tissues during development (clustering according to binary distance, Ward’s method). This analysis revealed a strong relationship between limb and facial tissue, also observed in clustering of specific histone modifications (Extended Data Fig. 4a), and further supporting the hypothesis of that facial structures and limbs have a common developmental origin^92,93. c, Enrichment of functional terms (x-axis, GO biological processes, P values from GREAT binomial test; FDR is Benjamini–Hochberg corrected q value) for the sets of strong enhancers (state no. 5) across each tissue-stage (y-axis). Sample sizes provided in Supplementary Table 5. The terms were hierarchically clustered (average linkage) according to Pearson’s correlation. A subset of the terms highly enriched in both limb and face is listed below the main heatmap. d, Fraction of bases (x-axis) annotated in the indicated state consistently in up to seven stages sampled (y-axis). Only tissues sampled at seven stages are shown here (n = 5). e, Sankey diagram showing the origin and fate of all genomic intervals classified as TSS-distal strong enhancers (state no. 5) in E14.5 forebrain. The chromatin state classification of these regions was tracked across the available developmental stages, and the relative genomic coverage of each chromatin state at each transition is plotted. The thickness of each colour (y-axis) indicates the coverage of each state.