The reproducibly-positive SNPs identified by abuser/control comparisons in both European- and African-American samples cluster together in small chromosomal regions (Fig. 2) to extents much greater than anticipated by chance. 937 of the reproducibly-positive SNPs from approach (1) lie in 271 clusters of ≥3 SNPs that are separated from each other by ≤25kb. This degree of clustering was never identified by chance (Monte Carlo p<0.00001). These clusters of reproducibly positive SNPs from analytic approach (1) identify 104 genes (Tables 1, 2). Randomly selected SNPs never cluster by chance within genes to the extent observed here (Monte Carlo p<0.00001).
