The SAGE EA dataset used in the present study was part of the meta-analysis of rs1229984 reported by our group (Bierut et al, 2011). In the current EA subset of SAGE we validated the reported association of rs1229984 with maxdrinks (p = 4.32 × 10−7). Although the signal was not genome-wide significant, it explained a large proportion (1%) of variance for maxdrinks in the SAGE EAs. This SNP was also nominally associated with maxdrinks in the COGA family dataset (p = 0.012) with protective effect (−0.19). The meta-analysis of these two datasets for rs1229984 resulted in genome-wide significant association with maxdrinks (p = 2.04 × 10−8). To test the possible moderating effect of rs1229984 on the relationship between maxdrinks and other genetic variants, we further analyzed the COGA and SAGE GWAS including rs1229984 as a covariate. Rs4758317 in LMO1 remained the strongest signal with a slight improvement in effect (beta value changed from −0.078 to −0.080) and p value (changed from 7.2 × 10−7 to 5.1 × 10−7). The meta-analysis resulted in a nominal improvement in the evidence for association of some other SNPs as well, but no new suggestive association results emerged through this analysis (supplementary table 4).
