Modeling the Effects of Genetic Drift and Natural Selection with Discrete Simulation
by
Spencer Tipping
Truman State University
Coauthors: Karen O'Connell (keo332@truman.edu) Dianne Kopp (dek453@truman.edu) Anton Weisstein (weisstae@truman.edu) Pam Ryan (pjryan@truman.edu)

To obtain distributions against which to compare our calculated Dsyn and Dnon values, we constructed a simulation that models the effects of selection and genetic drift over time. This software provides the capability to integrate and model several different evolutionary patterns such as natural selection, genetic drift, or any combination of these. A model-specific function determines the reproductive fitness of each individual, and a real-valued matrix controls both the degree of reproductive isolation of subpopulations and changes in population size. By combining these parameters, we can simulate both simple and more complex scenarios; for example, directional selection occurring within partially isolated subpopulations. After many model runs under a specific scenario, we acquire statistical distributions of our modified D statistics. Using Bayesian analysis, we then calculate the likelihood of the observed D statistics under each of the evolutionary scenarios to infer the population’s demographic and evolutionary history. Although our model was constructed to reflect specific biological features of HIV evolution, we anticipate that our extension of Tajima’s analysis will have broad applicability.

Date received: May 14, 2008