data and imputed where necessary6,7,30. We performed DNase-seq association mapping by regressing the adjusted sensitivity in each window against the genotypes at variants in a 40 kb region centered on each DHS. As validation, we used our DNase-seq reads as well as ChIP-seq reads and DNase-seq reads from ENCODE to confirm that allele-specific reads spanning heterozygous sites at dsQTLs are consistent with the association analysis. We also used RNA-seq data from the same cell lines8 to study the links between dsQTLs and eQTLs. Finally, we explored the properties of dsQTLs that make them more or less likely to influence gene expression by fitting a logistic model on all dsQTLs, where the eQTL status of each dsQTL-eQTL test is modeled as a function of distance from the TSS and a variety of other annotations. For full details of all methods see the Supplementary Information.
