for CHRNA5. We added additional SNPs to these bins, based on strong LD with tag SNPs in the larger ECLIPSE GWAS dataset. We also identified 3 sets of SNPs (3a, b, c in Table 4) with cis-eQTL associations for CHRNA5 and moderate LD with SNPs in bin 3 (r2 0.57–0.76). SNPs in bins 1 and 2, but not bin 3, showed evidence of association with COPD in the combined GWAS dataset, though they were not genomewide significant (bin 1: rs1051730, p = 2.8e-6; bin 2: rs938682, p = 5.6e-5; bin 3: rs6495306, p = 0.2). These results suggest that the COPD-associated SNP rs1051730 (bin 1) may influence phenotype by its effect on gene expression, while COPD-associated SNPs in bin 2 (tagged by rs938682) may exert their effect through other mechanisms. SNPs in bin 3, although eQTLs, were not associated with COPD risk.
