Next, we simulate genotypes at the disease SNP in each individual, where the prevalence is fixed at 0.1%. For a fixed high-risk allele frequency and heterozygous genotype relative risk, genotypes are generated for each case-control sample assuming Hardy-Weinberg equilibrium and a multiplicative disease model. In the same way as before, the disease SNP allele frequency is generated in each external cohort using the Balding-Nichols model. Given that the external samples are unselected with respect to the disease phenotype of interest, their genotypes are generated under the assumption of Hardy-Weinberg equilibrium, irrespective of the disease model.
