To quantify the degree to which multiple cis-associated SNPs comprising a given cis-eQTL represent multiple, independent effects, we applied a stepwise regression framework to each probe that had at least two significant cis-eQTL SNPs at the 0.01 permutation threshold. We identified a total of 33 genes with multiple eQTLs (0.15% of all 21,800 genes tested), corresponding to 1.1% of genes with significant cis-eQTLs, or 1.7% of genes overall that had more than two significant cis-eQTL SNPs in at least one population. In CEU, fourteen genes exhibited multiple independent cis-effects (corresponding to 1.8% of genes with significant cis-eQTLs). In total, 10 genes (1.2%), 1 (0.14%), 7 (0.83%), 1 (0.12%), 0 (0%), 7 (0.71%), and 13 (1.5%) with multiple cis-eQTLs were detected for CHB, GIH, JPT, LWK, MEX, MKK and YRI respectively (Table S6). Taken together, at the 0.01 permutation threshold for all eight populations, we observed that ∼0–2% of genes with an expression association possess multiple independent cis-eQTLs effects. At most, a single gene had five independently associated SNPs, i.e., expression of TACO1 (Syn CCDC44), translational activator of mitchondrially encoded cytochrome c oxidase subunit I, a gene with a role in Leigh Syndrome, was independently associated with five SNPs in LWK.
