Understanding Human Demography And Its Implications For The Detection And Dynamics Of Natural Selection
Patterns of genetic variation from living people can provide important information regarding ancient demographic history as well as how recombination and natural selection operate in different populations. However, extracting important information from large-scale datasets remains challenging. In this dissertation, I develop and validate statistical methods to understand human demography, patterns of recombination across the genome, how demography impacts the ability to detect positive selection, and how demography influences the proportion of deleterious genetic variation within a population. First, I develop a novel haplotype-based approach to estimate bottleneck parameters from human single-nucleotide polymorphism (SNP) data. Application of my method to simulated data shows it can reliably infer parameters from growth and bottleneck models, even in the presence of recombination hotspots when properly modeled. Application of the method to data collected by Perelgen Sciences shows evidence for a severe population size reduction in northwestern Europe starting 32,500- 47,500 years ago. Second, I compare patterns of linkage disequilibrium (LD) in HapMap data on the human autosomes to patterns to the human X chromosome. I find too little LD on the X chromosome relative to what is predicted under simple models based on the amount of autosomal LD. Third, I assess the effect of recent admixture on population genetic methods to infer ancient population growth. Haplotype methods are sensitive to recent admixture while methods based on SNP frequencies are less sensitive. Fourth, I evaluate the effect that recent admixture has on the ability to detect positive selection. Simulations show that admixture causes a decrease in power of some tests of selection, while increasing power for others. These results have important implications for detecting selective sweeps in admixed populations. Finally, I show that demographic history has had an important impact on patterns of segregating deleterious polymorphism in different populations. In particular, exon-resequencing data collected by Celera Genomics shows that European populations have a higher proportion of damaging mutations than African populations do. Through the use of forward-simulations with realistic demographic and selection parameters, I demonstrate that this pattern can be explained by the differing demographic histories of the two populations.
Dissertation or Thesis