Within each tissue, cis-eQTLs were identified by linear regression, as implemented in FastQTL71, adjusting for PEER factors, sex, genotyping platform, and three genotype-based principal components (PCs). We restricted our search to variants within 1 Mb of the TSS of each gene and, in the tissue of analysis, minor allele frequencies ≥0.01 with the minor allele observed in at least 10 samples. Nominal P values for each variant–gene pair were estimated using a two-tailed t-test. The significance of the most highly associated variant per gene was determined from empirical P values, extrapolated from a Beta distribution fitted to adaptive permutations with the setting –permute 1000 10000. These empirical P values were subsequently corrected for multiple testing across genes using Storey’s q value method16. To identify the list of all significant variant–gene pairs associated with eGenes, variants with a nominal P value below the gene-level threshold were considered significant and included in the final list of variant–gene pairs.
