We employed Spearman’s rank correlation to estimate the association strength between alleles of each SNP and gene expression (n=494), DNA methylation (n=468), and histone acetylation levels (n=433). We refer to the measurement unit of each molecular phenotype data as a feature and a significant association between a SNP and a feature as an xQTL (i.e. an xQTL is a SNP-feature pair). Based on the results of prior studies, we performed cis xQTL analysis between SNPs and each feature by defining a window size of 1Mb for eQTL analysis and haQTL analysis, and a 5Kb window for mQTL analysis21,22. The 1Mb window for haQTL analysis was motivated by the possibility that SNPs in enhancer regions, which are far away, can indeed impact gene regulation through interaction with promoter regions (e.g. chromatin looping). The much smaller window for the mQTL analysis was selected since the majority of cis mQTLs with the strongest correlation lie within a window of this size22. Also, the smaller window size helps reduce the multiple testing burden, given the much larger number of DNA methylation features.
