Figure 5.

Validation of GO Fish simulation results against δaδi. A complex demographic scenario was chosen as a test case to compare the GO Fish simulation against an already established site frequency spectrum (SFS) method, δaδi (Gutenkunst et al. 2009). The demographic model is from a δaδi code example for the Yoruba-Northern European (AF-EU) populations. Using δaδi parameterization to describe the model, the ancestral population, in mutation–selection equilibrium, undergoes an immediate expansion from N_ref to 2N_ref individuals. After time T₁ (= 0.005) the population splits into two with a constant, equivalent migration, m_EU-AF (= 1) between the now split populations. The second (EU) population undergoes a severe bottleneck of 0.05N_ref when splitting from the first (AF) population, followed by exponential growth over time T₂ (= 0.045) to size 5N_ref. The SFS (black dashed line) above is of weakly deleterious, codominant mutations (2N_refs = −2, h = 0.5) where 1001 samples were taken of the EU population. The spectrum was then normalized by the number of segregating sites. The corresponding GO Fish parameters for the evolutionary scenario, given a mutation rate of 1 × 10⁻⁹ per site, 2 × 10⁹ sites, and an initial population size, N_ref, of 10,000, are: T₁ = 0.005 × 2N_ref = 100 generations, T₂ = 900 generations, m_EU-AF = 1/(2N_ref) = 0.00005, 2N_refs = −4, h = 0.5, and F = 0. As in δaδi, the population size/time can be scaled together and the simulation will generate the same normalized spectra (Gutenkunst et al. 2009). Using the aforementioned parameters, a GO Fish simulation ends with ∼3 × 10⁶ mutations, of which ∼560,000 are sampled in the SFS. The red line reporting GO Fish results is the average of 50 such simulations; the dispersion of those 50 simulations is reported in Figure S1. Each simulation run on the NVIDIA GeForce GTX 980 took roughly the same time to generate the SFS as δaδi did [grid size = (110,120,130), time-step = 10⁻³] on the Intel Core i7 4771, just < 0.7 sec.