To control the PAR (population attributable risk) of each group, and ensure that all variants have a low effect, we sample each variant in a group using the same marginal PAR (α), so that α is the group-PAR divided by the number of disease-risk contributing variants (D-variants). Each variant is sampled independently. The mutation probability in unaffected individuals is sampled according to the frequency spectrum described above, and the genotype probabilities in unaffected individuals are calculated assuming Hardy-Weinberg proportions. The odds ratio (r) of each genotype is calculated from the genotype probability in the unaffected individuals (qU) usingand the genotype probability in the affected individuals (qA) is calculated asSee e.g. Ref. [36]. The population of affected and population of unaffected individuals are sampled using qA and qU respectively. We simulate nA = 1000 affected and nU = 1000 unaffected individuals unless otherwise stated.
