This proposal develops software tools, which use Markov Chain Monte Carlo (MCMC) maximum-likelihood methods to infer population parameters from genetic data. We focus specifically on inferring selection and on mapping both known traits and unknown selective influences to specific chromosomal regions. We will develop new techniques for the following: (1) Estimating the presence and strength of natural selection and the degree of dominance, including statistical tests to compare selection hypotheses. (2) Mapping the location of a measured trait, or of a selection effect, relative to markers on a haplotype. (3) Mining the large sample of genealogies produced by MCMC algorithms for information such as the location of recombination hotspots, the time of significant events such as disease-locus mutations, and the overall time distribution of migration, mutation and recombination events. (4) Improving performance of MCMC algorithms via better search strategies and use of multiple computers in parallel. (5) Incorporating analysis of serial samples (samples taken from a population at different times) in order to strengthen estimation of selection and population growth. We will create freely distributed software implementing these methods and will test them using real and simulated data.
Showing the most recent 10 out of 16 publications
