For these analyses, we identified “reproducibly-positive” SNPs that cluster in small genomic regions within genes (Table 1). These SNPs a) display t values with p<0.05 significance in both African- and European-American abuser vs control comparisons (we define this evidence for significant association in each of two independent samples as “reproducible”), b) cluster, so that at least three reproducibly-positive SNPs lie within 25kb of other reproducibly-positive SNPs, c) identify genes (Table 2, legend). The 104 genes in Table 2 are supported by reproducible clustered positive association data from the same SNPs in each of two samples. We identify subsets of these genes when we impose more stringent criteria for: a) ≤10,000 basepair distances between reproducibly-positive SNPs, and b) at least 4 nominally-positive SNPs per cluster (Table 1).
