A subset of 180 TOPMed GeneSTAR European Americans samples also had RNA-seq data generated using platelets. eQTL analysis was performed as previously described54. Here, formal Bayesian co-localization was performed using the coloc55–57 package in R for each of the 16 independent loci (Table 1) against all gene transcripts where there was at least one SNP with an eQTL p value p < 0.003125 (0.05/16) for the specific gene within 20 KB of the peak variant. This yielded 10 locus-gene pairs (Supplementary Table 4). coloc tests five mutually exclusive hypotheses: H0, no GWAS and no eQTL association; H1, association with GWAS, but no eQTL; H2, association with eQTL, but no GWAS; H3, eQTL and GWAS association, but with two independent causal variants; and H4, shared causal variants for both eQTL and GWAS. The main interest is to assess whether there is a shared causal variant between eQTL and GWAS (i.e., H4). The package provides five posterior probabilities for these hypotheses (PP0, PP1, PP2, PP3, and PP4) and PP4 of >75% is considered evidence of a colocalization of GWAS and eQTL. Posterior probabilities for individual variants were evaluated once PP4 was met.
Do you have any questions about this protocol?
Post your question to gather feedback from the community. We will also invite the authors of this article to respond.