CAM colloquium - Friday, October 27
3:30 p.m.
655 Rhodes Hall

Speaker: Scott Williamson, Biological Statistics and Computational Biology, Cornell

 

Title: "Numerical analysis as a tool for statistical inference in population genetics"

Abstract: Poisson Random Field (PRF) theory provides a powerful likelihood and Bayesian framework for characterizing the forces that act on genetic variation, including mutation, natural selection, genetic drift, and demographic history. The original PRF model, developed by Sawyer and Hartl (1992), made some fairly restrictive and unrealistic assumptions regarding population processes, including equal fitness effects among new mutations, genic (additive) selection, random mating, and constant population size. Numerical methods for solving partial differential equations can be used to relax these assumptions and generalize the PRF approach. Here I will describe this approach, including methods for estimating parameters and testing hypotheses relating to population growth, bottlenecks, and divergence, and methods for characterizing natural selection in an appropriate demographic context. I conclude by using these methods to explore demographic history and natural selection in human populations.

 

Refreshments at 4:30 in 657 Rhodes Hall.