TABLE 2.

Summary of simulation assumptions

Single-locus model RL	Multilocus model P
Genotypic structure
One disease-susceptibility locus on one chromosome. There are one wild-type (N) allele and k − 1 disease (S) alleles.	Multiple (L) disease susceptibility loci on different chromosomes, with the same maximum number of allelic states (k).
Selection model
Fitness of an individual follows either an additive or a recessive model. Fitness of an individual with genotype NN, NS, or SS is 1, 1 − s/2, 1 − s for an additive model and 1, 1, 1 − s for a recessive model.	Each DSL follows either an additive or a recessive fitness model as in the case of the single-locus model. The overall fitness of an individual with fitness gi, i = 1,…, L at each DSL follows either a multiplicative () or an additive () multilocus model.
Mutation model
k-allele (Jukes and Cantor 1969) model with k > 105 to approximate the infinite allele model. Given mutation rate μ, an allele will mutate to any other state with equal probability , regardless of its current allelic state.	k-allele model with smaller k (e.g., k = 200) to be closer to a bidirectional mutation model ( and ). The mutation rate may vary from locus to locus.
Recombination
NA	NA because all DSL are physically unlinked.
Demographic
Instant, linear, or exponential population growth model. N0 = 104, N1 = 106 or 107.	Constant population size at N = 104 or 105.
Population structure
Population may be split into equally sized subpopulations (m = 10 or 100) before population expansion. During population expansion, these subpopulations may evolve with or without migration.	NA
Migration
Cyclic stepping-stone model at rate 10−3.	NA
Penetrance
Indirectly modeled, see explanation in the text.	Indirectly modeled, see explanation in the text.