NSFDEB-NERC: Machine learning tools to discover balancing selection in genomes from spatial and temporal autocorrelations

理解为什么物种内的个体在遗传上如此多样化是进化生物学和遗传学的一个基本问题。这种个体遗传多样性及其原因对生物多样性保护、农业生物学和生物医学具有重要影响。平衡选择是一个随着时间的推移促进和保持遗传多样性的过程。然而，尽管有一些著名的例子，人们对最近或短暂的平衡选择知之甚少，可能是因为它的遗传线索很微妙，很难与其他适应性和非适应性过程留下的线索区分开来。检测基因组数据中的平衡选择由于技术问题而进一步复杂化，例如缺失或退化的DNA序列数据，这些数据没有被当前的方法所考虑。该项目的主要目标是通过基于人工智能的最新进展设计最先进的工具来应对这些挑战，这些工具为识别遗传数据中过去进化事件的信号提供了策略。这些工具将在一个公共储存库中免费提供，以便广泛使用。此外，该项目将通过iDeepLearn夏季研讨会积极吸引当地高中生参与编码和机器学习，并通过FAU校园高中的外展计划吸引其他STEM代表性不足的学生。总之，这些计划中的活动将促进我们对不同分类群体之间平衡选择的理解，并促进传统上代表性不足的高中生参与STEM研究。通过使用通常从古代DNA获得的时间采样遗传数据来增强检测平衡选择，这存在许多技术障碍。这项研究旨在开发新的机器和深度学习方法，可以从空间和时间采样的遗传数据中识别最近和短暂平衡选择的基因组特征，同时考虑研究人员使用古代DNA和非模型生物体时遇到的技术问题。该项目将专门解决从不完整、低质量、未分阶段或在遗传和人口参数不确定的情况下合并的数据中检测平衡选择的问题。开发的方法将应用于人类，蚊子和果蝇的试验台，因为这些研究系统有证据证明不同的平衡选择模式，以及公开可用的数据集，这些数据集具有项目寻求克服的技术障碍的特征。这些方法也将作为开源工具实施，适用于模型和非模型生物体中常见的各种数据类型，通过消除数据质量和人口统计学知识限制所带来的障碍，为未来的适应性研究提供支持，并最终导致对生命树适应性历史的理解的新见解。人口基因组学机器学习研讨会将作为佛罗里达大西洋大学iDeepLearn暑期项目的一部分，为高中女生开发和提供。在美国和英国，将开发和提供针对中学生的多个STEM相关职业活动。该奖项反映了NSF的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

Understanding why individuals within species are so genetically diverse is a fundamental problem in evolutionary biology and genetics. This individual genetic diversity, and its causes, has important consequences for biodiversity conservation, agricultural biology, and biomedicine. Balancing selection is a process that promotes and maintains genetic diversity over time. Despite a few well-known examples, however, little is known about recent or fleeting balancing selection, likely because its genetic clues are subtle and difficult to distinguish from those left by other adaptive and nonadaptive processes. Detecting balancing selection in genome data is further complicated by technical issues, such as missing or degraded DNA sequence data, which are not accounted for by current methods. The primary goal of this project is to tackle these challenges by designing state-of-the-art tools based on recent advances in artificial intelligence, which provide strategies for identifying signals of past evolutionary events in genetic data. These tools will be made freely available in a public repository, enabling widespread use. In addition, this project will actively engage local high school students in coding and machine learning through the iDeepLearn summer workshop, and other students from groups under-represented in STEM through outreach programs at the FAU campus high school. Together, these planned activities will facilitate future advancements in our understanding of balancing selection across diverse taxonomic groups, as well as foster participation of traditionally underrepresented high school students in STEM research. Detecting balancing selection is enhanced by using temporally sampled genetic data often accessed from ancient DNA, which presents numerous technical hurdles. This research seeks to develop novel machine- and deep-learning methods that can identify genomic signatures of recent and transient balancing selection from spatially and temporally sampled genetic data, while accounting for technical issues encountered by researchers working with ancient DNA and nonmodel organimsm. The project will specifically address detecting balancing selection from data that are incomplete, low-quality, unphased, or pooled under settings for which there is uncertainty in genetic and demographic parameters. Developed methods will be applied to human, mosquito, and fruit fly testbeds, as these study systems have evidence for diverse modes of balancing selection, and publicly available datasets with characteristics of the technical hurdles the projects seeks to overcome. These methods will also be implemented as open-source tools applicable to a wide range of data types common across model and nonmodel organisms, empowering future studies of adaptation by removing barriers imposed by limitations of data quality and demographic knowledge, and ultimately leading to novel insights in the understanding of adaptive history across the tree of life. Workshops on machine learning in population genomics will be developed and delivered for high school girls as part of the iDeepLearn summer program at Florida Atlantic University. In both the US and the UK, multiple STEM-related career events aimed at secondary school pupils will be developed and delivered.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.