Speaker
Jukka Corander, Professor, Department of Mathematics and statistics, University of Helsinki
Abstract
History of genetically diverged populations can be inferred from genotype data by modeling the process of neutral random changes in allele frequencies over time. We have recently introduced two such approaches (see here and here) where drift and mutation effects are considered using a diffusion-based approximation to the allele frequency distributions under the neutral Wright-Fisher and infinite alleles models. Inference of the population genealogies and divergence times under these models is computationally tractable even for large population samples, in contrast to a coalescent-based approach. We have performed inference using a combination of analytical and numerical techniques, including adaptive importance sampling and adaptive MCMC. The models could also adapted to screen for loci that have been under selection.