Novel Methods for Characterizing Recombination and Selection

重组和选择是影响基因组变异模式的两种主要进化机制。推断历史重组模式的努力是疾病关联研究设计和分析的核心，识别选择目标的能力可能对生物医学研究具有重要意义。这一应用的长期目标是定量描述重组和选择对基因组变异的影响。为了在群体基因组学中进行准确的推断，将开发高效的算法和严谨的数学技术。这项应用的具体目标是：目标1：开发评估蒙特卡罗方法的方法，以进行重组合并中的似然计算。确定性的、基于算法的方法将被开发来非常准确地计算可能性，为测试和微调蒙特卡罗方法的可能性计算开辟了新的机会之窗。对于中等大小的输入数据，新开发的工具将用于评估现有的蒙特卡罗方法。目的2：发展表征历史交叉和基因转化重组的方法。为了获得精确的多点条件抽样分布，将建立一个基于扩散近似的通用数学框架。使用这种方法，将开发出一种可以联合估计交叉和基因转化率的方法。此外，将重新审查现有的估计方法，并将进行具体的计算改进。目的3：研究自然选择在多个基因座上的相互作用。对多个位点的选择交互作用进行了解析研究，并对交互选择形成的LD结构进行了表征。还将研究多基因座选择下的固定概率。相关性：了解人类基因组的变异模式对于研究疾病风险的遗传基础和药物反应的可变性至关重要。这项研究的目的是开发准确的方法来表征塑造基因组变异模式的各种进化机制。

Recombination and selection are two major evolutionary mechanisms that influence the pattern of variation in genomes. Efforts to deduce patterns of historical recombination are central to the design and analysis of disease association studies, and the ability to identify targets of selection may have important implications for biomedical research. The long-term objective of this application is to characterize quantitatively the effects of recombination and selection on genomic variation. Efficient algorithms and rigorous mathematical techniques will be developed for accurate inference in population genomics. The specific aims of this application are: Aim 1: Develop methods to assess Monte Carlo approaches to likelihood computations in the coalescent with recombination. Deterministic, algorithm-based methods will be developed to compute likelihoods very accurately, opening up a new window of opportunities for testing and fine-tuning Monte Carlo approaches to likelihood computation. For input data of moderate size, the newly developed tool will be used to evaluate existing Monte Carlo methods. Aim 2: Develop methods to characterize historical crossover and gene-conversion recombinations. A general mathematical framework based on diffusion approximation will be developed to obtain accurate multi-locus conditional sampling distributions. Using that approach, a method that can jointly estimate crossover and gene-conversion rates will be developed. Further, existing estimation methods will be revisited and specific computational improvements will be made. : Aim 3: Study the interaction of natural selection at multiple loci. The interaction of selection at multiple loci will be studied analytically and the structure of LD shaped by interacting selection will be characterized. Fixation probabilities under multi-locus selection will also be studied. ¿ Relevance: Understanding the pattern of variation in the human genome is central to the study of the genetic basis of disease risk and variability in drug response. The aim of this research is to develop accurate methods to characterize various evolutionary mechanisms that shape the pattern of genomic variation.