Although we are encouraged by our findings, we recognize that there are limitations to our study. Among our top candidate genes, we only detected nominal association for C15orf53 and SC in SAGE. Several reasons could possibly explain this limited replication. First, the power to replicate findings of small effect across studies in samples of the size used in this study is low. Second, in contrast to the COGA sample, neither the SAGE nor the OZALC samples were ascertained from large families severely affected by AUDs. It is possible that severely affected families have a concentration of genetic variants that influence risk for alcoholism and that may have less effect on alcoholism in the general population. A coordinated evaluation including many more families severely affected by alcoholism is necessary to confirm our findings.
