In order to replicate our observations from the GABRIEL data set, we repeated the analysis using the results from the EVE meta-analysis, an asthma study incorporating results from GWAS conducted in several ethnically diverse North American populations (32). We note that the studies included in the EVE meta-analysis were entirely independent from those included in the GABRIEL meta-analysis. We used the results for combined meta-analysis incorporating all populations. Because the EVE study imputed SNPs using HapMap data, this resulted in the inclusion of 2 340 251 SNPs in our replication efforts. We proceeded by extending the same integration approach detailed for the lymphocyte count and GABRIEL data to the EVE results [i.e. calculating the median P-value for our candidate genes, allowing for proxy SNPs to replace unrepresented eQTLs (Supplementary Material, Fig. S7C), and assessing significance by permutation testing]. Five of the 10 239 genes were not represented by a proxy SNP in the EVE data set and were therefore excluded from the replication test.
