Characterizing the Genetic Structure of Populations

Abstract

The ability to characterize the genetic structure of populations continues to be central to all aspects of population and quantitative genetics, and measures based on correlations of pairs of alleles continue to be as relevant now as they were when discussed by Wright in 1951.The need to develop methodology for the characterization of the genetic structure of natural populations, especially of humans, has increased with recent large-scale disease association studies. We have proposed various estimators for population structure measure Fst, such as the moment estimator,the Kappa type estimator and the likelihood estimator. The likelihood estimates based on assumed normality of allele frequencies over populations have proven difficult to obtain in practice. Maximum likelihood estimates based on the Dirichlet distribution will provide a more robust framework, in spite of the implicit assumptions for that distribution not applying to microsatellite markers. The behavior of moment and MLE estimates in different scenarios were explored. Another population specific Fst estimator was proposed using GEE2 method. The proposed method yielded consistent results which are demonstrated via simulations.