Methods and Software for Large-Scale Gene-Environment Interaction Studies

项目摘要/摘要 复杂的人类疾病和相关的数量性状是许多风险因素的相互作用，包括遗传 和环境成分。基因-环境相互作用研究是一个可以使用的一般框架 确定改变环境、生理、生活方式或治疗效果的遗传变异，以及 那些在复杂特征上造成年龄、性别、种族/民族差异的人。此外，遗传关联研究 对基因与环境的相互作用进行了解释，以加深我们对基因 通过允许在不同的暴露层中产生不同的遗传效应，来确定复杂疾病的结构。与 最近技术的进步和成本的降低，基因和基因组数据正在以非常大的规模产生 比例。然而，用于基因-环境相互作用研究的常用统计软件程序有 一般是在很多年前开发的，它们的计算算法还没有优化来分析 来自可能复杂的研究设计的数十万到数百万个样本。填补两国之间的空白 大规模基因-环境相互作用研究的当前和未来分析需求以及当前的分析 作为解决方案，我们计划(目标1)开发针对常见变异基因-环境相互作用的高效算法 分析规模与样本量呈线性关系；(目标2)开发罕见变异基因的新统计检验- 相关样本的混合效应模型框架中的环境相互作用分析；和(目标3) 在开源新软件程序中实施所提出的统计方法和计算算法。 我们的目标1解决了目前在进行基因-环境相互作用研究中的计算挑战 多达数百万个样本。在目标2中，我们计划解决基因环境中的统计和计算挑战 大规模全基因组测序数据的相互作用分析，说明相关性、复杂性研究 设计，以及错误的型号说明。目标3专注于软件开发，我们将很好地交付- 用于大规模基因环境的文件化和用户友好的软件包和分析管道 互动研究。这些方法和软件程序将应用于正在进行的全基因组测序 项目以及生物库规模的数据，它们将极大地促进大规模遗传和 未来几年基因-环境相互作用研究的基因组数据，以更好地了解遗传基础 复杂的心脏代谢、肺、血液、睡眠疾病及其年龄、性别、种族/民族差异，以及 在精准健康研究中推广个性化疾病防治策略。

PROJECT SUMMARY/ABSTRACT Complex human diseases and related quantitative traits are the interplay of many risk factors, including genetic and environmental components. Gene-environment interaction studies are a general framework that can be used to identify genetic variations that modify environmental, physiological, lifestyle, or treatment effects, as well as those contributing to age, sex, racial/ethnic disparities on complex traits. Moreover, genetic association studies accounting for gene-environment interactions are conducted to enhance our understandings on the genetic architecture of complex diseases by allowing for different genetic effects in different exposure strata. With the recent advances in technology and lowering costs, genetic and genomic data are being generated on very large scales. However, commonly used statistical software programs for gene-environment interaction studies were generally developed many years ago, and their computational algorithms have not been optimized to analyze hundreds of thousands to millions of samples from possibly complex study designs. To fill in the gap between current and future analytical needs in large-scale gene-environment interaction studies and current analytical solutions, we plan to (Aim 1) develop efficient algorithms for common variant gene-environment interaction analyses that scale linearly with the sample size; (Aim 2) develop new statistical tests for rare variant gene- environment interaction analyses, in the mixed effects model framework for correlated samples; and (Aim 3) implement proposed statistical methods and computational algorithms in open-source new software programs. Our Aim 1 addresses current computational challenges in conducting gene-environment interaction studies in up to millions of samples. In Aim 2, we plan to solve statistical and computational challenges in gene-environment interaction analyses of large-scale whole genome sequencing data, accounting for relatedness, complex study designs, as well as model misspecification. Aim 3 focuses on software development and we will deliver well- documented and user-friendly software packages and analysis pipelines for large-scale gene-environment interaction studies. The methods and software programs will be applied to ongoing whole genome sequencing projects, as well as biobank-scale data, and they will significantly facilitate the use of large-scale genetic and genomic data for gene-environment interaction studies in upcoming years to better understand the genetic basis of complex cardio-metabolic, lung, blood, sleep diseases and their age, sex, racial/ethnic disparities, and promote personalized disease prevention and treatment strategies in precision health research.