CAREER: New Statistical Approaches for Studying Evolutionary Processes: Inference, Attribution and Computation

该奖项的全部或部分资金来自《2021年美国救援计划法案》(公法117-2)。从DNA等分子序列样本中进行统计推断提出了一系列基本挑战。这些挑战包括对样本的祖先和过去的进化史进行复杂的建模，以及大量和嘈杂的数据。遗传数据的持续大规模增加导致了目前的方法不适用于现有数据量的情况，研究人员被迫对现有数据进行缩减抽样或从不充分的汇总统计数据中推断参数。这项研究项目将解决从现代分子数据中推断最佳设计的聚合体模型的需求。聚合体是家谱的概率模型，也就是代表样本祖先的树。联合模型用于推断具有科学相关性的参数，如有效种群数量、迁徙模式和选择。这个项目的研究目标是扩大合并模型的类别，并设计新的高效统计算法，使我们能够解决许多促进科学发展的实际问题。此外，这些项目的成果将促进新的统计理论和可处理的方法的发展，有助于生物解决方案。该项目还概述了一项广泛的教育和外联活动的积极计划，这些活动将扩大对统计科学的参与，并将加强科学方面更具包容性的氛围。参与该项目的本科生和研究生将获得一个独特的机会，在统计科学和生物学的界面上进行跨学科的动手研究培训，使他们能够为进化生物学、分子生物学、种群遗传学、系统发生学、癌症基因组学、概率建模、统计推断和相关领域的进步做出贡献。PI将积极参与多项外联活动，如斯坦福大学本科生暑期研究计划，该计划将允许招募更多样化的未来数据科学家池，并在科学领域培养更具包容性的气候。该项目的研究成果将作为统计遗传学新课程的基础，并将整合到本科生和研究生课程中。具体地说，该项目将扩展合并模型的类别，并通过利用谱系的度量概念、马尔可夫链的可集总性和分而治之的策略来提供一套新的算法和统计方法。具体目标包括(1)开发合并模型，以纳入各种抽样方案和生物学过程，例如动态种群结构、重组和强选择；(2)为合并理论和应用开发一个度量框架；(3)为进化参数的贝叶斯推断开发可扩展的策略；(4)实施、验证和分析传染性疾病的分子序列，例如SARS-CoV-2、古代和现代人类DNA样本以及癌症单细胞变异。此外，该项目将积极促进在多个战线扩大对统计科学的参与，从基于团队的跨学科研究培训到社区外展。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2). Statistical inference from a sample of molecular sequences such as DNA poses a series of fundamental challenges. These challenges include complex modeling of the sample's ancestry and past evolutionary history, large and noisy data. The ongoing large-scale increase of genetic data has led to a situation in which current methods are not applicable to the amount of data available and researchers are forced to down-sample available data or to infer parameters from insufficient summary statistics. This research project will address the need for optimally designed coalescent modeling for inference from modern molecular data. The coalescent is a probability model on genealogies, that is, the trees which represent the ancestry of the sample. Coalescent models are used for inferring parameters of scientific relevance such as effective population size, migration patterns and selection. The research goals of this project are to expand the class of coalescent models and to design novel efficient statistical algorithms, allowing us to address many practical problems that advance science. Furthermore, the outcomes of the projects will foster the development of new statistical theory and tractable methods that contribute to biological solutions. This project also outlines an active plan for a broad range of educational and outreach activities that will broaden participation in statistical sciences and will enhance more inclusive atmosphere in science. The undergraduate and graduate students involved into the project will be offered a unique opportunity for interdisciplinary hands-on research training at the interface of statistical sciences and biology, allowing them to contribute to progress in evolutionary biology, molecular biology, population genetics, phylogenetics, cancer genomics, probabilistic modeling, statistical inference, and related fields. The PI will actively participate in multiple outreach activities such as the Stanford undergraduate summer research program, which will allow for recruiting more diverse pool of future data scientists and for fostering more inclusive climate in science. The research findings of the project will serve as foundation for new program in statistical genetics and will be integrated into undergraduate and graduate courses. Concretely, this project will expand the class of coalescent models and provide a suite of new algorithmic and statistical approaches by exploiting a metric notion of genealogies, lumpability of Markov chains and divide-and-conquer strategies. The specific aims include (1) develop coalescent models to incorporate various sampling schemes and biological processes such as dynamic population structures, recombination and strong selection; (2) develop a metric framework for coalescent theory and applications; (3) develop scalable strategies for Bayesian inference of evolutionary parameters and (4) implement, validate and analyze molecular sequences of infectious disease such as SARS-CoV-2, ancient and modern human DNA samples and cancer single cell variation. Furthermore, the project will actively contribute to broadening participation in statistical sciences at multiple fronts, from team-based interdisciplinary research training to community outreach.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.