For active enhancer-associated H3K27ac peaks (available in 98 cell types), we found a similar number of enriched studies (47 at 2% FDR, Extended Data 12b), but for promoter-associated H3K4me3 and H3K9ac peaks, we found only 25 and 18 enriched studies, respectively (Extended Data 12a,b), suggesting that enhancer-associated marks are more informative for tissue-specific disease enrichments than promoter-associated marks. For DNase peaks, we only found 9 enriched studies (Extended Data 12c), partly because they were only available in 53 reference epigenomes (restricting H3K4me1 to the same 53 resulted in 25 enriched studies, Table S6), and possibly due to lack of distinction between enhancer and promoter regions. For transcription-associated H3K36me3, we found 15 enriched studies (Extended Data 12d), indicating that these help capture additional biologically-meaningful variants outside annotated promoter and enhancer regions. In contrast, we found no enriched study for either Polycomb-associated H3K27me3 peaks or heterochromatin-associated H3K9me3 peaks (Extended Data 12e,f). These results indicate that enhancer-associated marks have the greatest ability to distinguish tissue-specific enrichments for regulatory regions, but promoter-, open-chromatin-, and transcription-association marks are also significantly enriched, indicating that disease variants affect a wide range of processes.
