For ANYDEP, rs1229984 in ADH1B, the most well-replicated signal for alcohol dependence12 was not GWS. Even relative to findings from the companion paper by Lai et al., where rs1229984 was GWS in the EA+AA analysis (beta=−0.86, p=1.72E-8), the association between rs1229984 and ANYDEP was considerably weaker both in magnitude and significance (Supplemental Table S7), despite a considerably larger analytic sample. Additional loci identified for alcohol dependence diagnosis, symptom count and individual criteria in Lai et al., also did not achieve GWS for ANYDEP, although all signals were nominally significant with effects in an identical direction. An important distinction between these two companion studies is noteworthy. While Lai et al., excluded individuals with ≥ 2 alcohol or drug abuse/dependence criteria from their unaffected population, the current study allowed these individuals to remain in the unaffected group. Thus, variations across findings in the two studies might be due, not only to differing definitions of affecteds, but also the definition of unaffecteds. Finally, the current study did not identify the same SNPs as were noted in our prior study of ANYDEP and its
