of closely genetically related individuals (i.e., with an estimated genetic relatedness larger than 0.025) was dropped, which left for the analysis 3616 distantly related individuals. We specified the prevalence as equal to 22 %, value chosen in line with the prevalence of cannabis use estimated in Europeans (European Monitoring Centre for Drugs and Drug Addiction, 2012). Heritability of age at onset was not estimated as GCTA cannot handle survival data. We also investigated the relationship between chromosome length and the amount of variance explained in the trait. Consistent with the model of a polygenic trait, we expect—on average—the longer chromosomes to explain a larger amount of the variance. We tested this in a linear regression (one-tailed test) where we regressed the estimated proportion of variance explained by each chromosome on the chromosome length.
