In the present study, after merging 480 COGA subjects into the SAGE sample, most results were consistent with a previous study that used the SAGE sample alone (Bierut et al., 2010). The top-ranked risk SNPs (p<10−5) in European-Americans, African-Americans, and the combined samples in that previous study (Bierut et al., 2010) were confirmed by our analysis (summarized in Zuo et al., 2012). In the present study, we found three top-ranked risk genes (p<5×10−7 together with FDR<0.05) for alcohol dependence (SERINC2, KIAA0040 and IPO11) in European-American discovery sample. Two of these genes, i.e., KIAA0040 and IPO11, were also among the top-ranked genes in the European-Americans as reported previously (Bierut et al., 2010). However, the most significant gene (SERINC2) in the present study has not been reported before. It has been neglected because it was only suggestively-significant in the separate SAGE and COGA datasets in previous GWASs. Here, by combining both datasets to increase statistical power (with overlapping excluded), stringently cleaning the phenotype and genotype data and controlling for various confounding effects, we discovered this risk gene locus. This locus was specific
