In our assessment of allele frequency differences between abusers and controls 83,202 and 75,327 SNPs displayed “nominally positive” t values with p<0.05 in African- and European-American samples, respectively. There was substantial convergence of the results from these two GWA datasets using the non-template (1) “converge then cluster” GWA analysis approach. 11,037 of the 870,000 tested SNPs displayed “reproducible” results, as defined using this approach. These SNPs thus displayed nominally significant abuser vs control allele frequency differences in each of the two samples. This overall convergence was much greater than anticipated based on chance. None of 100,000 Monte Carlo simulation trials that each began by selecting 83,202 and 75,327 random SNPs displayed as many as 11,037 nominally significant results in both samples (p<0.00001). None of 10,000 permutation trials displayed results from permuted datasets that matched or exceeded the 11,037 SNPs actually observed (p<0.0001 by permutation analyses). These 11,037 SNPs thus provide the “reproducibly positive SNPs” for analytic approach (1).
