One way of bypassing the sample size issue is to study intermediate phenotypes in population isolates that are more genetically and environmentally homogeneous. The first published GWAS on resting EEG power, an intermediate phenotype for alcoholism, was performed in a sample of 322 Plains American Indians [76]. Although the diagnostic phenotype of alcoholism did not generate genome-wide significant statistical signals, the EEG GWAS identified three genes, one of which (SGIP1) was associated with alcoholism, an effect that might be mediated via the same brain mechanisms accessed by EEG power. Convergence of findings at the sub-threshold level with previous findings in genetic studies of addictions suggested that the intermediate phenotype approach can potentially identify genes that have a general effect on addiction even in datasets of modest size, notably of population isolates [76].
