Novel Statistical Methods for Development of Polygenic Scores in Multi-Ancestry Cohorts

项目摘要 多基因评分(PGSS)是衡量一个人患疾病的遗传风险的指标，根据以下结果得出 全基因组关联研究(GWAS)汇总统计1。它们是一种很有前途的工具，可以在 高遗传风险，也可用于评估风险因素的因果效应和检查基因环境 互动2.然而，PGS高度依赖于祖先，并且当前的PGSS在 统计遗传研究中代表性不足的群体3。需要开发新的PGS方法，该方法 可以改善这些代表不足的人群的PGS表现。特别是，改善了心房的PGS 房颤(AF)对于确保所有人都能获得房颤基因研究的进展至关重要。房颤经常是 无症状且未经治疗可导致其他心血管疾病，包括心力衰竭、中风、 和心血管疾病死亡率4。有了改进的房颤PGS，高危个体可以被识别和治疗。 以前的工作已经证明，跨祖先合并GWAS结果可以改善PGS 性能，然而这项工作仅限于两个祖先6，7。 使用多祖先数据构建PGSS以提高PGSS的性能，特别是在 代表性不足的人群。我们将采取两种不同的方法结合多重祖先的结果- 具体的全球气候变化数据。我们的第一个目标是开发一种方法来创建改进的特定于祖先的分数。我们的第二个 AIM将开发一种方法来创建一个跨祖先PGS。我们将评估我们新的 使用模拟研究的统计方法，并将使用数以百万计的AF数据来验证我们的方法 退伍军人计划(MVP)8.此外，我们将把我们的新方法提供给更多的研究人员 社区通过在GitHub上发布我们的方法。我们将把这些方法的应用重点放在自动对焦上，但我们的 方法论可用于多种疾病。 改进PGS方法，以便它们对基因表达不足的个人表现良好 为了确保基因研究的进步对所有人都有利，研究势在必行。我的指导团队已经 在房颤的基因研究方面有出色的经验，并致力于支持我的训练和 职业发展。我们设计了一项培训计划，其中包括机制培训和 心血管疾病流行病学、高级统计方法和职业发展 例如科学写作和负责任的研究行为。通过这一奖学金，我将发展技能，以 实现我的长期目标，成为一名统计遗传学的独立研究员，拥有以下专业知识 心血管疾病。

Project Abstract Polygenic scores (PGSs) are measures of an individual’s genetic risk of disease, derived from the results of genome-wide association study (GWAS) summary statistics1. They are a promising tool to identify individuals at high genetic risk and can also be used to assess causal effects of risk factors and examine gene-environment interactions2. However, PGSs are highly ancestry-dependent, and current PGSs do not perform well in underrepresented populations in statistical genetic research3. There is a need to develop new PGS methods that can improve PGS performance in these underrepresented populations. In particular, improved PGS for atrial fibrillation (AF) is vital for ensuring that advances in genetic research of AF are available to all. AF is often asymptomatic, and without treatment can lead to other cardiovascular diseases, including heart failure, stroke, and cardiovascular mortality4. With improved PGS for AF, high risk individuals can be identified and treated5. Previous work has demonstrated that combining GWAS results across ancestries can improve PGS performance, however this work has been limited to two ancestries6, 7. We propose to develop methods for constructing PGSs with multiple ancestry data to improve the performance of PGSs, in particular in underrepresented populations. We will take two distinct approaches combining the results of multiple ancestry- specific GWAS data. Our first aim will develop a method to create improved ancestry-specific scores. Our second aim will develop a method to create one trans-ancestry PGS. We will assess the performance of our new statistical methods using simulation studies, and will validate our methods using AF data from the Million Veterans Program (MVP)8. Additionally, we will make our novel methods available to the greater research community by publishing our methods on GitHub. We will focus our applications of the methods on AF, but our methods can be used for a wide range of diseases. Advancing PGS methods so that they perform well for individuals who are under-represented in genetic studies is imperative for ensuring that advances in genetic research are beneficial to all. My mentoring team has outstanding experience in genetic research of AF, and is committed to supporting me in my training and professional development. We have designed a training plan which includes training in mechanisms and epidemiology of cardiovascular disease, advanced statistical methodologies, and professional development such as scientific writing and responsible conduct of research. Through this fellowship, I will develop the skills to achieve my long-term goal of becoming an independent researcher in statistical genetics with expertise in cardiovascular disease.