Replication was defined as the query variant (original lead or proxy) having a nominal eQTL P < 2.2x10-04 for the same eGene (corresponding to P = 0.01 / 45, where 45 refers to the total number of tissues tested). A lenient threshold was chosen in order to rule out any evidence of replication. High-LD proxies for a lead were defined as having r2 > 0.8 in the UK10K European reference panel and located in the same cis window. All available LD-proxies were tested for replication and the variant with the most significant eQTL P-value in the discovery tissue was stored as the proxy. If no LD-proxies could be tested, the cis variant with the most significant eQTL P-value in the discovery tissue overall was selected. Overall, the same lead variant was available to test in 95% of tests across tissues (median of discovery tissues, with each discovery tissue represented as the median of tests across all replication tissues; 90% for iPSC eQTLs). A high-LD proxy for the lead was tested 3.6% of the time (7.3% in iPSC), while the best
