Like with other complex traits, our results demonstrate that substance use phenotypes are polygenetic and moderately to highly heritable. Using standard biometric twin models, heritabilities ranged from 43% for Alcohol Consumption to 58% for Behavioral Disinhibition (Figure 1). Additive SNP effects estimated by GCTA on the parent sample account for 16% of the variance in Alcohol Dependence to 22% of the variance in Drug Use. While the aggregate additive SNP effect, estimated by GCTA is relatively large (e.g., 10–30% of total phenotypic variance; Figure 2), identifying and summing actual individual SNPs with genome-wide scoring yields much weaker effects, accounting for around 0.25% of the variance in the substance use and behavioral disinhibition measures. Dividing the GCTA results estimated in the parent sample by the twin-estimated heritabilities allows us to estimate the total heritable variance accounted for by the aggregate SNP effect from GCTA. We estimate that the additive SNP effect accounts for 21% (Alcohol Dependence), 32% (Behavioral Disinhibition), 36% (Nicotine Use/Dependence), 38% (Alcohol Consumption), and 45% (Drug Dependence) of the heritable variance in these traits.
