There was limited replication and overlap for the top signals identified in either the COGA or the SAGE dataset (Supplementary Tables 1 & 2). The association of SNPs (p < 1 × 10−4) near the neural cell adhesion molecule (NCAM2) and M-phase phosphoprotein (MPHOSPH6) genes observed in the COGA study showed nominal association in the SAGE study (0.04 ≤ p ≤ 0.05). The association of SNPs near the LIM domain only (LMO1), Toll like receptors (TLR1, TLR10) and Ataxin 2 binding protein 1 (A2BP1) genes identified in SAGE were also weakly (0.03≤p≤0.06) associated with maxdrinks in COGA. However, the number of SNPs with p < 1 × 10−4 in both COGA and SAGE with the same direction of effect in the other study was significantly higher than expected by chance: 270/479 of the COGA SNPs (P=0.003) and 380/646 of the SAGE SNPs (p=4×10−6).
