Although we believe those non-replicable GWAS findings have lower possibility to be true than the replicable findings, we do not completely exclude the possibility that those non-replicable risk genes might harbor causal variants for alcohol dependence in some specific populations or for some specific endophenotypes of alcohol dependence. The non-replicability might be attributable to the heterogeneity of phenotype or population of different samples. More homogenous samples may be helpful with reducing the false negatives in this non-replicability.
