Statistical Methods for Genetic Risk Predictions across Diverse Populations

摘要 尽管全基因组关联研究在识别遗传变异方面取得了非常成功的成果 与复杂的疾病和特征有关，将GWAS的结果转化为临床仍然具有挑战性 有用的疾病风险模型，用于改进疾病预测、预防、诊断、预测、监测和 治疗。此外，到目前为止，大多数GWA都集中在欧洲血统的个人身上， 这使得在其他人群中推导风险模型变得困难。最近的研究表明，共有的基因 跨人群对复杂疾病的贡献以及考虑功能性疾病的潜在好处 跨种群分析中的注解。该项目的最终目标是开发严谨、高效、 以及稳健的综合建模方法，通过利用海量数据进行跨人群的风险预测 大量公开可用的GWAS摘要数据，丰富的功能注释，以及越来越多的 研究对象为来自代表性不足人群的参与者。这将通过五个具体的 目标。前三个目标将为跨人群风险预测开发三种互补的方法， 包括：(目标1)贝叶斯方法(ME-PRED)，沿着我们发表的工作的路线，将其中之一 功能注释信息或多个特性信息，从明确建模联合效果大小 多种群和函数注释；(目标2)考虑了 更通用、更灵活的效应大小分布和不需要验证队列的统计推断 用于调整某些模型参数；以及(目标3)一种快速且稳健的贝叶斯非参数方法(SDPR)，该方法 对不同的遗传结构具有高度的适应性，并且计算效率很高。广泛的模拟将 以比较这些方法和其他现有方法的性能。在目标4中，我们将 应用这些方法来评估各种疾病和特征的潜在临床实用性，重点是 代表性不足的人群。我们还将与耶鲁世代项目的调查人员密切合作 研究这些工具对其研究参与者的潜在好处，包括许多代表不足的人 人口。然后，我们将改进一些方法的实现，以减少计算时间和 改进了AIM 5中的用户界面和分析管道。我们已经组建了一个调查团队， 在统计遗传学、医学遗传学和高性能计算方面的专业知识以开发、实施、 评估和传播所建议的方法。如果成功，这些方法和工具将带来更多 在代表性不足的人群中进行准确的遗传风险预测，满足减少健康的关键需求 贫富差距。

Summary Although genome-wide association studies (GWAS) have been very successful in identifying genetic variants associated with complex diseases and traits, it is still challenging to translate GWAS results into clinically useful disease risk models for improved disease prediction, prevention, diagnosis, prognosis, monitoring, and treatment. Furthermore, most GWAS conducted to date have focused on individuals of European ancestry, making it difficult to derive risk models in other populations. Recent research has suggested shared genetic contributions to complex diseases across populations and the potential benefit of considering functional annotations in cross-population analysis. The ultimate objective of this project is to develop rigorous, efficient, and robust integrative modeling approaches for risk prediction across populations by capitalizing on the vast amount of publicly available GWAS summary data, abundant functional annotations, and a growing number of studies with participants from underrepresented populations. This will be accomplished through five specific aims. The first three aims will develop three complementary approaches for cross-population risk predictions, including: (Aim 1) a Bayesian approach (ME-Pred), along the line of our published work to incorporate either functional annotation information or multiple trait information, that explicitly models joint effect sizes from multiple populations and functional annotations; (Aim 2) an empirical Bayes approach (GWEB) that considers a more general and flexible effect size distribution and statistical inference that does not need a validation cohort for tuning some model parameters; and (Aim 3) a fast and robust Bayesian nonparametric method (SDPR) that is highly adaptive to different genetic architectures and is computationally efficient. Extensive simulations will be performed to compare the performance of these methods and other existing methods. In Aim 4, we will apply these methods to evaluate the potential clinical utility for various diseases and traits, with a focus on underrepresented populations. We will also work closely with investigators from the Yale Generations Project to study the potential benefit of these tools for its study participants, including many from the underrepresented populations. We will then refine the implementations of some methods to reduce computational time and improve the user interface and analysis pipeline in Aim 5. We have assembled a team of investigators with expertise in statistical genetics, medical genetics, and high-performance computing to develop, implement, evaluate, and disseminate the proposed methods. If successful, these methods and tools will lead to more accurate genetic risk predictions in underrepresented populations, addressing a critical need in reducing health disparity.