Leveraging functional data to predict disease risk in multi-ethnic populations

项目摘要 全基因组关联研究在识别遗传变异方面取得了广泛的成功。 与常见疾病风险相关，导致在欧洲人群中成功预测疾病风险 使用多基因风险分数。不幸的是，多基因风险评分的准确性存在很大差距。 欧洲人和非欧洲人之间的差异，以便临床努力改善生物医学结果 通过精准医疗可能会加剧健康差距。多种族数据的可用性不断提高 更大的样本量提供了提高多基因风险评分的准确性的机会，通过改善 因果变异的定位和具有群体特异性效应的变异的辅助鉴定。值得注意的是， 功能基因组学数据具有改进所有这些努力的巨大潜力，但还不够充分 融入多民族方法。在这里，我们建议获得综合分析的优势 多种族和功能数据，建立在我们的疾病研究计划广泛进展的基础上 在过去的8年里，我们在多民族人口中进行了测绘；我们目前应用的重点是适应 现有的统计方法在新的背景下，整合了多种族和功能的数据，这是目前遭受的 在可用方法上存在很大差距。我们的研究将由来自2,500,000个多民族的经验数据推动 样本(&gt；1,500,000个具有基因型/表型数据，&gt；1,000,000个具有汇总关联统计数据)， 包括非洲裔美国人、拉丁美洲人、东亚人和南亚人的样本，这些样本跨越了一系列疾病 和数量表型。我们将分析个人级别的数据和摘要级别的数据，并 整合功能数据集，包括全基因组功能注释和基因表达数据。

Project Summary Genome-wide association studies (GWAS) have been broadly successful in identifying genetic variants associated to common disease risk, leading to successes in predicting disease risk in European populations using polygenic risk scores. Unfortunately, there is a large gap in the accuracy of polygenic risk scores between European and non-European populations, such that clinical efforts to improve biomedical outcomes via precision medicine may exacerbate health disparities. The increasing availability of multi-ethnic data in larger sample sizes provides opportunities to improve the accuracy of polygenic risk scores, by improving localization of causal variants and aiding identification of variants with population-specific effects. Notably, functional genomics data has great potential to improve all of these efforts, but has yet to be adequately integrated into multi-ethnic approaches. Here, we propose to reap the advantages of integrative analyses of multi-ethnic and functional data, building on the extensive progress of our research program on disease mapping in multi-ethnic populations over the past 8 years; the focus of our current application is on adapting existing statistical methods to a new setting, integrating multi-ethnic and functional data, which currently suffers a large gap in available methods. Our research will be driven by empirical data from >2,500,000 multi-ethnic samples (>1,500,000 with genotype/phenotype data and >1,000,000 with summary association statistics), including African American, Latino, East Asian and South Asian samples spanning a wide range of diseases and quantitative phenotypes. We will analyze both individual-level data and summary-level data and incorporate functional data sets, including genome-wide functional annotations and gene expression data.