Genetic & Social Determinants of Health: Center for Admixture Science and Technology

项目摘要 必须了解个人之间巨大健康差异的根本原因， 生活在美国的不同种族和族裔群体。遗传学和基因之间的复杂关系 社会因素影响健康结果。在美国，大约33%的人属于少数民族。 约12.5%的人生活在联邦贫困线以下。历史上和最近的欧洲人，土著人的混合 美国人，非洲人和亚洲人导致美国人口有相对大量的混合 这些人的祖先来自他们自我认同的种族之外。我们所有人（AoU）计划和百万 退伍军人计划（MVP）包括所有参与者的遗传，健康和社会经济信息， 因此，提供了一个机会，以确定造成健康差距的因素。然而，AoU计划 和MVP要求他们的数据保留在本地托管站点内，因此对这些队列进行联合分析 需要开发能够实现隐私保护分布式计算的算法（即，没有 显示个人层面的数据）。在理解健康的遗传决定因素方面存在三个重要差距： 1)大多数研究都是由欧洲人主导的，虽然他们控制了全球血统，但在欧洲， 没有试图模拟混合个体的地方祖先特征的拼凑; 2）GWAS是 主要使用SNP进行，而祖先特异性遗传变异（串联重复序列）的重要来源 (TRs)和主要组织相容性复合体（MHC）间隔）不被测定;和3）大多数GWAS不 调整社会经济因素。美国医学遗传学和基因组学学院（ACMG） 发表了一份医学上可操作的癌症和心血管基因清单，建议将偶然的 发现致病性变异，以降低发病率和死亡率，但少数群体被排除在医疗保健之外 由于共同的障碍（例如，语言和访问）使得很难区分基因 和社会经济因素导致不同的健康结果。CAST（Center for 混合物科学和技术）计划是为了提高临床实用的遗传信息的所有 生活在美国的人口。在目标1中，我们将开发和应用疾病风险预测的多变量模型 包括当地祖先、复杂变体（TR和HLA类型）。在目标2中，我们将进行可扩展的 分布式计算使用来自AoU和MVP计算飞地中数百万个人的数据。在Aim中 3，我们将开发新的方法，利用电子健康记录和调查， AoU和MVP，评估在我们的模型中包括健康的社会决定因素的影响，并前瞻性地 与新的AoU和MVP参与者一起评估它们。为了实现这些目标，我们组织了一个高度 跨学科的研究小组，在遗传学，基因组生物学，数据共享政策和 技术，健康差异，表型和统计。

PROJECT SUMMARY It is imperative to understand the underlying sources of the large health disparities among individuals from different racial and ethnic groups living in the United States (US). Complex relationships between genetics and social factors influence health outcomes. Approximately 33% of people in the US belong to an ethnic minority group and ~12.5% live below the federal poverty line. Historical and recent mixing of Europeans, Native Americans, Africans and Asians resulted in the US population having a relatively large number of admixed individuals who carry ancestry from outside their self-identified race. The All of Us (AoU) Program and the Million Veterans Program (MVP) include genetic, health and socioeconomic information on all participants, and therefore provide an opportunity to identify factors contributing to health disparities. However, the AoU program and MVP require their data to stay within local hosting sites, therefore conducting joint analyses on these cohorts requires the development of algorithms that enable privacy-protecting distributed computing (i.e., without revealing individual-level data). There are three important gaps in understanding genetic determinants of health: 1) most studies have been dominated by European individuals, and while they control for global ancestry, there is no attempt to model the patchwork of local ancestry characteristic of admixed individuals; 2) GWAS are primarily conducted using SNPs, while important sources of ancestry-specific genetic variation (tandem repeats (TRs) and the major histocompatibility complex (MHC) interval) are not assayed; and 3) most GWAS do not adjust for socioeconomic factors. The American College of Medical Genetics and Genomics (ACMG) has published a list of medically actionable cancer and cardiovascular genes recommended for return of incidental findings of pathogenic variants to reduce morbidity and mortality, but having minorities excluded from healthcare follow up due to common barriers (e.g., language and access) makes it difficult to distinguish between the genetic and socioeconomic factors that contribute to disparate health outcomes. The goal of the CAST (Center for Admixture Science and Technology) program is to improve the clinical utility of genetic information for all populations living in the US. In Aim 1, we will develop and apply multivariate models of disease risk prediction that incorporate local ancestry, complex variants (TRs and HLA types). In Aim 2, we will conduct scalable distributed computing using data from millions of individuals across the AoU and MVP compute enclaves. In Aim 3, we will develop new approaches to characterize phenotypes using electronic health records and surveys from AoU and MVP, assess the impact of including social determinants of health in our models, and prospectively evaluate them with new AoU and MVP participants. To achieve these goals, we assembled a highly interdisciplinary group of researchers with expertise in Genetics, Genome Biology, Data Sharing Policy and Technology, Health Disparities, Phenotyping, and Statistics.