We computed the VEGAS summary statistic and associated p-value for 17,567 genes for each of the five phenotypes. For none of the phenotypes did the p-value for any gene reach significance at the Bonferroni corrected threshold of .05/17,567 = 2.8 × 10-6. The smallest gene based p-value we observed was 6.6 × 10-6 for the association of Nicotine with DMRTB1, a transcription factor. We also sought to determine whether there was enrichment of low p-values for 165 autosomal genes identified as being candidates for substance abuse and related phenotypes in previous research (Hodgkinson et al., 2008; S. F. Saccone et al., 2009). Full results for these candidate genes are listed in Supplemental Table 6. For Alcohol consumption, 4.8% of the 165 candidate genes were associated at p < .05 versus 5.0% of non-candidate genes. The comparable percentages of gene-based tests meeting the nominal p < .05 threshold for candidate genes versus genes not in the candidate set were: 6.1% versus 5.4% for Alcohol Dependence; 6.7% versus 7.3% for Illicit Drugs; 4.2% versus 5.7% for Nicotine; and 3.6% versus 4.7% for Behavioral Disinhibition. None of the phenotypes showed significant enrichment of candidate genes at the nominal p-value threshold of .05.
