GWAS studies (Figure 5A, Table S6). Next, we hypothesized that SNPs shared between xQTL types, which affect multiple molecular phenotypes, are more likely to impact downstream processes and could constitute a list of “high confidence” functional SNPs. We therefore compared all xQTL SNPs that are shared across at least two molecular traits, against those xQTLs that are only found for one molecular trait. Indeed, we observed enrichment for the shared xQTLs, but their enrichment was not always higher than the background xQTL SNPs, i.e. somewhat trait dependent (Table S6). To test the robustness of the results to window size, we repeated the analysis with 100Kb windows for all three xQTL types (Table S7). The overall trend remained the same with slightly higher enrichment observed.
