Before combining the UHS drug abuse cases in each ancestry group with population controls of the same ancestry, we needed to ensure that there were no underlying biases driven by any one of the control data sets. To check for such bias, we made pairwise comparisons between the control data sets based on overlapping genotypes and arbitrary case assignments to subjects from one data set and control assignments to subjects from the other data set. Using SNP association results with the arbitrary case/control assignments, genomic control inflation factors (λgc) were computed. Control data sets used in the final analysis data set had λgc≤1.05 for all pairwise comparisons (see Supplemental Information, Tables S6 and S7).
