Two Problems in Population Genetics

Genome-wide association (GWA) studies conducted in large human cohorts have revealed that many traits are highly polygenic, with numerous loci throughout the genome contributing small additive effects. These findings imply a simple prediction model, often referred to as a polygenic score, in which an individual's predicted phenotype is the inner product of their genotype and a set of weights corresponding to the effects of every site in the genome. It is now widely appreciated that the accuracy of a focal individual's polygenic score depends on their genetic relationship to the population in which the prediction model was generated. Genetic predictions are usually less accurate for more distantly related, out-of- sample individuals. When environmental conditions are constant, this reduction in accuracy can largely be attributed to differences in allele frequencies and patterns of linkage disequilibria between the GWA study population and the population from which the focal individual was sampled. The primary aim of the first project is to determine what proportion of reduced accuracy can be attributed to the former—referred to as allelic turnover—in isolation. In Chapter 2, I develop a theoretical framework to investigate the effects of allelic turnover on the accuracy of out-of-sample polygenic scores. The second thrust of my thesis describes the distribution of mitotic loss of heterozygosity (LOH) throughout the genome of the plant pathogen, Phytophthora capsici. In Chapter 3, I develop a procedure to infer LOH events simultaneously in multiple members of a clonal lineage. As isolates within a clonal lineage differ only by mutations accrued during mitotis, LOH events identified in lineages can be more readily attributed to mitotic processes, distinct from meiotic processes, such as inbreeding, which also produce runs of homozygosity. I apply the method to two large genotyping-by-sequencing data sets of P. capsici.