COMPOSITE LIKELIHOOD AND REGRESSION BASED METHODS FOR INFERRING POPULATION GENETIC PARAMETERS FROM DNA SEQUENCES

This doctoral dissertation is composed of three projects and four chapters. Chapter 1 presents the background theories and models in the field of population genetics that are related to these projects. In Chapter 2, I develop a composite likelihood ratio test (CLRT) for detecting genes and genomic regions that are subject to recurrent natural selection while relaxing the assumption of free recombination. We find that the test has excellent power to detect weak negative selection and moderate power to detect positive selection. Moreover, the test is quite robust to the bias in the estimate of local recombination rate, but not to certain demographic scenarios such as population growth or a recent bottleneck. In Chapter 3, I present a novel method, Poisson pairwise difference method (PPDM), which efficiently co-estimates the selection coefficient γ = 4Nes and mutation rate θ = 4Neµ from arbitrarily correlated SFS data. We demonstrate that the PPDM log-likelihood ratio test has good power to detect positive selection and moderate power to detect weak negative selection