Fig. 2

Distal enhancer landscapes of GC cell lines. a Histone profiles of OCUM1 and SNU16 cells show enrichment of H3K27ac and H3K4me3 around the UTP15 TSS. A predicted distal enhancer enriched for H3K27ac and > 2.5 kb distant from UTP15 TSS is observed. b Comparison of H3K27ac signals over common predicted enhancers between two KATOIII replicates. c Genome-wide average profile of chromatin marks (H3K27ac, H3K4me1, and H3K4me3) and DNA methylation (5mC) at all predicted enhancers and active promoters. Active promoters are those annotated promoters overlapping H3K27ac peaks. H3K27ac, H3K4me1, and H3K4me3 profiles were generated by ChIP-seq. 5mC profiles were generated by MeDIP-seq. RPM: Reads per million mapped reads. The “summit” of a predicted enhancer region refers to the midpoint of the bimodal peak. The indicated windows (1 kb, 6 kb) were chosen as indicated to highlight the bimodal pattern of histone marks. d Recurrence rates of predicted enhancers. Recurrent predicted enhancers were identified as those enhancers occurring in at least two GC cell lines. Data presented are the mean percentage +/− standard deviation of commonly predicted enhancers found in two or more gastric cancer cell lines, as a function of the number of cell lines. e Distribution of predicted super-enhancer and typical enhancers present in GCs across the human genome. f t-distributed stochastic neighbor embedding (t-SNE) analysis using predicted enhancers present in GCs reveals separation between GCs and matched normal tissues (n =36). g Difference in somatic point mutation rates among predicted super-enhancers, typical enhancers, and randomly selected genomic regions. P value: two-side Student’s t test. Relative somatic point mutation rate was calculated as the log2 fold change of the somatic point mutation rate over the background point mutation rate