We created a hypothetical case-control exome sequencing study involving real sequencing data and simulated phenotype data using 438 individuals, split in two populations (see Methods). To induce population stratification, we assigned case-control status to each sample randomly with a bias to take more cases from one population, and more controls from the other population. Association tests indicated inflated rates of spurious statistically significant p-values. We corrected for population stratification by modifying the permutation scheme to account for subpopulations. This correction was effective at controlling Type-I errors in all association tests.
