One other study has utilized results from TAG to predict cannabis involvement in an independent sample. Vink and colleagues (2014) used polygenic scores created by aggregating SNPs at varying p-value thresholds, up to p=0.5, to predict tobacco, alcohol and cannabis phenotypes in an independent Dutch sample. Such an approach has the advantage of utilizing effects, even those that are not statistically significant, to create a genetic predictor that is more consistent with the polygenic view of complex behavioral traits. The TAG polygenic scores for CPD were associated with CPD (p=0.012 for TAG SNPs with p-value < 0.5) and cannabis use (p<.001) but the extent to which the most significant SNPs (e.g. rs1051730) contributed to these findings is unclear. Our analyses studied these highly significant individual SNPs that were part of the polygenic scores and while we note a similar degree of association with rs1051730, we do not see any association with other TAG SNPs (e.g. rs3733829). We also extended the association to cannabis abuse/dependence, a phenotype that was not studied by Vink et al. (2014)
