Testing all markers 10Mb surrounding SERINC2 in the European-American discovery sample, we found that all association signals with p<10−4 were concentrated in a narrow region around SERINC2 (Figure 1A), within which all association signals with p<10−3 were concentrated in a 246Kb NKAIN1-SERINC2 region (Figure 1B). This region contained five genes including NKAIN1, SNRNP40, ZCCHC17, FABP3 and SERINC2; full names and biological functions of these genes and their associations with human diseases are summarized in Table S610. Among all of the 97 unimputed SNPs in this region, 74 SNPs were nominally associated with alcohol dependence (1.7×10−7≤p≤0.026) in the European-American discovery sample (data not shown). These 74 risk markers were located in three linkage disequilibrium (LD) blocks that were defined by Gabriel et al. (2002) using the program Haploview (Figure 1I). Imputing across the NKAIN1-SERINC2 region, we found that among all 795 SNPs, 471 SNPs were nominally associated with alcohol dependence (1.7×10−7≤p≤0.047), 53 of which survived region-wide and cohort-wide correction for multiple testing (corrected α=1.7×10−5). The risk alleles of all 471 markers were minor alleles (f<0.5). The top-ranked 27 SNPs (p<10−6; Table
