Extended Data Fig. 2: Characterisation of dynamically active enhancers.

a, Chromatin profile heatmaps of ATAC, H3K4me3, H3K4me1, H3K27ac, H3K27me3 and input signals in hPGCLCs at ATAC-seq summit ± 3 kb. Segregation of ATAC-seq summits by K-means clustering using normalised chromatin mark signals. b, Distribution of chromatin mark signals at active, mixed, primed, poised, repressed, and neutral enhancers in hPGCLCs (see Fig. 2a). Enhancers per violin/box plot: 23255 active, 1288 mixed, 36999 primed, 5648 poised, 3984 repressed, 79290 neutral. Box plots depict the median, lower and upper hinges correspond to the 25th and 75th percentiles. c, Enhancer state transitions of DE-active enhancers. Distal OCRs not overlapping any histone modification peak in the analysed cell types were referred to as ‘neutral’ enhancers. d, Putative enhancers with differential H3K27ac levels (absolute(log2(fold change)) >1 and adjusted p-value < 0.05) in the indicated sample pairs. e, High confidence enhancer-gene associations. Putative enhancers were assigned to the nearest TSS. The relevance of the enhancer-gene pair was assessed by the Kendall’s rank correlation analysis between the enhancer H3K27ac signals and the expression levels of the associated genes across the 6 cell types and 2 replicates (see Methods). In the simplified model shown, the Enh1-geneA and Enh3-geneB pairs (green text and arrows) were identified as high confidence associations based on positive correlation between H3K27ac and gene expression levels. f, Expression levels of genes associated with active enhancers in different cell types. Compared to simply associating genes to the nearest active enhancer, high confidence active enhancer associated genes (Kendall rank correlation coefficient >0.3; empirical p-value < 0.05) exhibited significantly higher expression in all cell types. Two-sided Wilcoxon rank sum test with FDR correction. Gene number per violin/box plot (all nearest, high confidence): hESC (8686, 1279), PreME (9782, 1413), ME (11494, 1853), DE (12279, 2224), hPGCLC (9954, 1726), hPGC (6612, 1024). Box plot organisation as in Extended Data Fig. 2b. g, Distribution of ATAC, H3K27ac and H3K27me3 signals in dynamically active enhancers and high confidence target gene expression. Enhancers were segregated into nine clusters (Fig. 2c). Enhancer per clusters as in Fig. 2c. Box plot organisation as in Extended Data Fig. 2b.