The genome-wide association statistics deviated slightly from the null (λGC=1.092) (Supplementary Figure 1), though the LD score regression intercept of 1.001 suggested that the inflation of the test statistic was due to polygenicity rather than any population stratification.35 Genome-wide significant associations (P<5 × 10−8) were obtained for 14 loci after clump-based pruning (Table 1 and Figure 1). These included 8 hits on chromosome 4q23/4q22.3 that span ~2.3 MB and several alcohol dehydrogenase genes (ADH1B, ADH1C, ADH5) that have all previously been implicated in alcohol consumption.9, 12 Conditional SNP GWAS of the variants on chromosome 4q suggests that there are 4 independent hits in this region: rs145452708 in the ADH1B/ADH1C region, rs29001570 in ADH5, rs35081954 in ADH1C and rs193099203 that lies in an intergenic region of 4q23. Most of the associated variants on 4q23 are relatively rate (MAF 0.006–0.01) with the minor allele associated with lower alcohol consumption. However, rs35081954 is a common insertion/deletion polymorphism with a MAF of 0.42. Two hits were identified on chromosome 4p14 in the KLB gene that was previously associated with alcohol consumption in a large
