Statistical modeling of genetic histories and relationships of populations.

摘要

Five statistical explorations of the inference of population histories are described. Chapter 1 considers genetic diversity, relationships, past expansions in size, and divergence times of eleven human populations, as inferred from population allele frequencies at microsatellite loci in two different data sets. This chapter determines which of three population growth statistics are most efficient and demonstrates that cladograms constructed from genetic distance matrices do indeed improve when larger data sets are used. Chapter 2 is a case study of microsatellite genotypes in eight human populations, as well as a demonstration of the use of individual genotypes in population structure inference. This study shows that the historical record of isolation of the Libyan Jewish population agrees with the microsatellite profile of this group. Chapter 3 is an evaluation of methods of population structure determination, using a data set taken from twenty chicken breeds. Individual clustering algorithms are found to be highly effective in population structure analysis, and as few as 12–15 loci and 10–15 individuals from each hypothesized population are sufficient to accurately identify breeds of origin of individuals. Chapter 4 calculates the numerical difference between the coalescence time of allelic copies taken from two populations and the divergence time of the populations. If no migration takes place between populations after divergence, the coalescence time exceeds the divergence time. If migration has taken place between the populations, coalescence is more recent than divergence with high probability. Chapter 5 explores the probability of topological concordance of gene trees and species trees when samples of arbitrary size are taken from the species. An expression for the probability of topological concordance is given in the three-species model. For most species divergences, the topological concordance probability is found to reach its limiting value at reasonably small sample sizes.