The genotyped participants’ ancestral proportions were determined using the STRUCTURE program (Pritchard et al., 2000) with reference to HapMap populations. Following exclusions for European Americans having >25% African ancestry and African Americans having <25% African ancestry, the heroin addiction case and control datasets closely resembled the expected ancestral proportions (Figure 1). Participants were also excluded for call rate <90%, sample duplication, gender discordance, excessive homozygosity, and first-degree relatedness. Our final analysis dataset included 7,095 European Americans (711 cases and 6,384 controls) and 3,662 African Americans (1,293 cases and 2,369 controls), as shown in Table 1.
