We also evaluated associations of top SNP findings from prior GWAS of specific drugs, including heroin/opioids,(Gelernter et al., 2013; Liu et al., 2013; Nielsen et al., 2010) cocaine,(Gelernter et al., 2014b) and marijuana(Agrawal et al., 2014) Results for 96 of the 127 previously reported SNPs are shown in Table S14. The other 31 SNPs were not available due to MAF<1% in either the 1000 Genomes AFR or EUR panels. We found nominally significant associations (meta-analysis P<0.05) for three SNPs previously implicated for marijuana: rs9891146 (a missense SNP in C17orf58, P=0.0059), rs17552189 (an intronic SNP in FLJ30838, P=0.020), and rs2381356 (an intronic SNP in the kynureninase [KYNU] gene, P=0.040): see Table S14 for details. Their directions of association were consistent with the prior GWAS of marijuana use disorder (Agrawal et al., 2014) The rs9891146 association was driven by EAs, mimicking its prior implication in this ancestry group; the associations for rs17552189 and rs2381356, which were originally implicated in a combined analysis of EAs and AAs, were strongest in AAs.
