We used genomic annotations based on DLPFC tissue from ChromHMM31 and computed the log odds of an xQTL SNP belonging to 1 of 15 regulatory regions (annotated by chromatin states) as compared to all non-xQTL SNPs proximal to molecular features, i.e. within 1Mb, 5Kb, and 1Mb windows for eQTL, mQTL, and haQTL analyses with all SNPs tested in these analyses considered as proximal. As shown in Figure 3A, eQTL SNPs are mainly enriched in promoters and transcribed regions, conforming to our understanding of how SNPs at transcription factor (TF) binding sites can affect protein-DNA interactions32 and how SNPs in transcribed regions are known to affect mRNA processing and turnover33. haQTL SNPs are also largely enriched in promoter and transcribed regions, consistent with the role of H3K9Ac in transcriptional activation34. By contrast, mQTL SNPs are mainly enriched in bivalent regions (promoters and enhancers) and PolyComb repressed regions, which matches prior findings that a large portion of mQTL SNPs resides in chromatin regions that are developmentally regulated22. Also, suppressed gene expression in PolyComb repressed regions might partly explain why eQTL and haQTL
