The additive genetic variance due to all SNPs (h2SNP) was similar across DD, PU, and DV. The univariate h2SNP estimate of DD was 0.36 (SE = 0.13, p-value (p) = 2.30E-3). The observed univariate h2SNP estimates for PU (h2SNP = 0.25, SE = 0.13, p = 0.03) and DV (h2SNP = 0.32, SE = 0.13, p = 8.91E-3) were slightly smaller. Consistent with the hypothesis that a large number of alleles throughout the genome contribute to substance dependence, longer chromosomes accounted for more phenotypic variation than smaller ones (Figure 2), but not for all phenotypes. For PU h2SNP increased with chromosome length (Figure 2; R2=0.27, β=1.87×10-4, t(20)=2.75, p=0.01). This was not the case with DD (R2=0.12, β=1.47×10-4, t(20)=1.64, p=0.12) and DV (R2=0.00, β=1.60×10-5, t(20)=0.20, p=0.84) phenotypes. Post hoc analyses of the h2SNP estimates within each cohort of SAGE (see Supplemental Table S3) showed internal consistency in the GCTA estimates for each of the phenotypes; though the small sample sizes and differences in ascertainment strategies across studies complicate direct comparison of these estimates.
