The Population Genetics of Divergence

描述(申请人提供)：种群分化，包括物种形成和种群结构的起源，是导致生命多样性的基本进化过程。这项研究项目将扩展基于可能性的发散模型方法的最新进展。新方法在马尔科夫链蒙特卡罗框架内对模型参数的先验分布进行解析积分。该方法得到一个联合概率密度函数，该函数与可用于嵌套人口统计模型的参数估计和对数似然比检验的似然比成正比。新方法将适用于发散中的一般多种群问题。长期以来，人们一直认为这类问题既需要种群遗传学的观点，也需要系统发生学的观点。研究计划概述了如何在通用的MCMC模拟下将这两者结合在一起。这将是第一个这样的方法，它不假设特定的系统发育；不假设基因流动没有发生；也不假设采样或祖先种群的相对种群规模。通过提供与似然成比例的联合后验密度的估计，该方法将提供直接进入对数似然比检验和基于似然的置信度区间的途径。该方法还将扩展到样本识别和DNA条形码的问题。这些新方法将被应用于黑猩猩物种和亚种之间差异的案例研究。这些方法也将应用于来自人类群体的大型多种群多位点数据集。

DESCRIPTION (provided by applicant): Population divergence, including speciation and the origins of population structure, is the fundamental evolutionary process leading to the diversity of life. This research project will extend recent advances in a likelihood-based approach to divergence models. The new approach employs analytic integration over prior distributions of model parameters within a Markov chain Monte Carlo framework. The method leads to a joint probability density function, proportional to the likelihood that can be used for parameter estimation and log- likelihood ratio tests of nested demographic models. The new method will be adapted to general multi-population problems in divergence. Such problems have long been appreciated as requiring both a population genetic perspective and a phylogenetic perspective. The research plan outlines how these two can be brought together under a common MCMC simulation. This will be the first such method that does not assume a given phylogeny; that does not assume that gene flow has not occurred; and that makes no assumptions about the relative population sizes of sampled or ancestral populations. By providing estimates of the joint posterior density, proportional to the likelihood, the method will provide direct access to log-likelihood ratio tests and to likelihood-based confidence intervals. The approach will also be extended to problems in sample identification and DNA barcoding. These new methods will be applied to a case study of divergence among species and subspecies of Chimpanzee. The methods will also be applied to large multi-population multi-locus data sets from human populations.