This broad tissue sampling affects the use of eQTL data for the interpretation of GWAS variants. We observed that a GWAS variant of interest is likely to be a cis-eQTL by chance. Of all common variants assayed within GTEx, 92.7% are nominally associated with the expression of one or more genes in one or more tissues (P < 0.05) and nearly 50% are significant when correcting for the number of tissues tested (Fig. 5b). Furthermore, linking an eQTL signal to a specific gene becomes increasingly complicated with abundant eQTL data. Some variants are associated with more than 30 different nearby genes (Extended Data Fig. 15a). Furthermore, even restricting to strong associations (P <10−10 in each tissue), for over 10% of eVariants, the gene with the strongest association varies between tissues (Fig. 5c; Extended Data Fig. 15b, c). These results reinforce the need for caution when using eQTL data to interpret the function of GWAS variants.
