genotypes and gene expression levels were determined by linear regression on quantile normalized gene-level expression values, after correction for known and inferred technical covariates (fig. S8) (14), using Matrix eQTL (21). To obtain gene-specific significance levels while correcting for testing multiple SNPs per gene, we computed permutation-adjusted P values for each gene for the most significant SNP per gene (14). We defined “eGenes” as genes with at least one SNP in cis significantly associated, at a false discovery rate (FDR) of ≤0.05, with expression differences of that gene (box S1) (14). A list of the significant SNP-gene pairs detected per eGene can be found on the GTEx Portal (http://gtexportal.org).
