Population Genetics of Polymorphism and Divergence for Diploid Selection Models With Arbitrary Dominance

摘要: We develop a Poisson random-field model of polymorphism and divergence that allows arbitrary dominance relations in diploid context. This provides maximum-likelihood framework for estimating both selection parameters new mutations using information on the frequency spectrum sequence polymorphisms. is first DNA sequence-based estimator parameter. Our also leads to likelihood-ratio test distinguishing nongenic from genic selection; simulations indicate this quite powerful when large number segregating sites are available. use explore bias parameter estimates caused by unacknowledged relations. When inference based polymorphisms, can be very strongly biased even minor deviations model. Surprisingly, however, (McDonald-Kreitman) data, nearly unbiased, completely dominant or recessive mutations. Further, we find weak overdominant increase, rather than decrease, substitution rate relative levels polymorphism. nonintuitive result has major implications interpretation several popular tests neutrality.