ABI Innovation: Coalescence-based Inference of Adaptation

该项目将开发一个数学框架，用于研究对局部环境压力反应不同的基因组序列。这一框架将比较在不同地点采集的样本的突变模式，以估计这些样本的可能谱系；这些谱系将被用于估计一些群体遗传模型的参数。这些参数将能够告诉我们地理位置之间潜在的选择差异。这一新的严格的数学框架有望取代目前临时和不充分的汇总统计数据。这一框架的潜在应用包括改进对疾病的干预(例如，针对艾滋病毒患者的个性化反应)，以及提高我们对哪些基因区域负责在恶劣环境中长期生存的理解。该框架将在独立的计算机软件中公开使用，这些软件可以在小型计算机或大型计算集群上运行。这项研究来自多个科学和技术学科(生物学、计算科学和统计学)，因此将提供一个很好的基础，在此基础上指导本科生、研究生和博士后，并培养他们对这个迫切需要更多培训机会的领域的兴趣。这个框架的基石植根于聚合理论，这是理论种群遗传学的一个分支，讨论个体的谱系形状，以及使用马尔可夫链蒙特卡罗技术评估不同情景和整合可能的解决方案的贝叶斯统计。这些数据将是已知包含技术错误的基因组序列；为了成功区分为特定环境选择的基因区域，必须考虑这些错误，但目前没有考虑这些错误。此外，来自不同地理位置(例如，不同的患者、不同的岛屿或不同的栖息地)的样本可以以不同的方式分组，这要求框架能够提供有助于对不同场景进行排序的最佳标准。进展和最终工作将记录在http://popgen.sc.fsu.edu和http://peterbeerli.com.网站上

This project will develop a mathematical framework for investigating genomic sequences that respond differently to local environmental stresses. This framework will compare mutation patterns from samples taken at different locations to estimate possible genealogies of these samples; these genealogies will be used to estimate the parameters of a number of population genetic models. These parameters will be able to tell us about potential selection differences among geographical locations. This new mathematically rigorous framework has the promise to supersede current ad hoc and inadequate summary statistics. Potential applications of this framework include improving interventions for diseases (e.g. individualized responses for HIV patients), and improving our understanding of which gene regions are responsible for long-term survival in harsh environments. The framework will be publicly available in standalone computer software that can be run on small computers or large computing clusters. This research draws from multiple science and technology disciplines (biology, computational science, and statistics) and thus will provide a great basis on which to mentor undergraduate, graduate, and postdoctoral students and foster their interest towards a field that desperately needs more training opportunities. The building blocks for this framework are rooted in coalescence theory, a branch of theoretical population genetics discussing the shapes of genealogies of individuals, and Bayesian statistics evaluating different scenarios and integrating over possible solutions using Markov chain Monte Carlo technology. The data will be genomic sequences which are known to contain technical errors; to successfully differentiate among gene regions that are under selection for particular environments, these errors must be taken into account, but currently are not. Additionally, samples from different geographical locations (for example different patients, different islands, or different habitats) can be grouped in different ways, which requires that the framework be capable of delivering optimality criteria that help to order different scenarios. Progress and the final work will be documented on the websites http://popgen.sc.fsu.edu and http://peterbeerli.com.