Improved detection of gene-diet interactions via longitudinal data, metabolomic proxies, and polygenic scores

摘要 饮食是代谢性疾病发展的关键因素，但饮食因素对疾病风险的影响是不同的 在个人之间。基因-饮食相互作用(GxD)可以帮助解决饮食反应中的这种个体间差异 通过识别改变饮食行为和代谢风险因素之间关联的基因变异 (例如，血糖性状和腰围)。识别许多这样的相互作用将使基因组引导 精确的营养建议(例如，地中海式饮食(MedDiet)与低脂饮食模式 降低糖尿病风险)。然而，自我报告的饮食行为中的不精确和偏见以及 变异体特有的效应在很大程度上阻碍了GxD相互作用的确定。这两把钥匙 障碍可以通过使用营养流行病学和基因组学的技术来改善GxD 身份证明。具体地说，饮食测量可以通过纳入(1)纵向饮食来改进 测量更稳定的摄入量估计和(2)饮食补充的目标代谢物 自我报告的数据。通过调整多基因风险分数(Prs)，可以改善效应规模小的问题。 创建多基因交互作用分数(IPR)的方法，该方法结合了多个变种的交互影响。 这些策略可以在Trans-Omics for Precision Medicine(TOPMed)队列中进行探索，该队列具有 多个时间点的自我报告饮食和血清代谢组学。MedDiet是一种非常适合的曝光方式 探索这些方法的方法：它通常被消费，影响代谢性疾病的风险，有证据表明 GxD相互作用，并具有经过验证的代谢组特征。有待探索的假设是，改进后的 膳食测量方法和IPR将允许复制已知和发现新的GxD。这个 首席研究员肯尼斯·韦斯特曼博士处于独特的地位，能够完成这项工作，作为一名早期- 具有营养背景、GxD分析专业知识和TOPMed膳食经验的职业科学家 数据。他需要额外的培训，以解决目前在纵向数据分析、代谢组学、 和领导力，促进完成以下研究目标。目标1：改善 结合纵向因素对代谢危险因素标准基因-饮食交互作用模型的重复性 测量和代谢性饮食替代物。目标2：改进交互测试的标准PRS 通过开发一种聚合遗传交互效应的多基因评分方法，目标3：发现 新的遗传因素改变了地中海式饮食和新陈代谢风险因素之间的联系。 拟议的研究将验证用于GxD研究的改进的饮食建模方法，并创建基因 预测饮食反应的分数。相关的培训将帮助韦斯特曼博士在 精准营养领域，并在未来的拨款中在饮食试验中进一步探索这些想法。

SUMMARY Diet is a critical factor in the development of metabolic diseases, but dietary factors affect disease risk differently across individuals. Gene-diet interactions (GxDs) can help resolve this inter-individual variability in diet response by identifying genetic variants that modify the association between dietary behaviors and metabolic risk factors (e.g., glycemic traits and waist circumference). Identifying many such interactions would enable genome-guided precision nutrition recommendations (e.g., Mediterranean-style diet (MedDiet) vs. lower-fat dietary pattern to decrease diabetes risk). However, imprecision and bias in self-reported dietary behaviors as well as small variant-specific effects have largely impeded the identification of robust GxD interactions. These two key obstacles can be addressed by using techniques from nutritional epidemiology and genomics to improve GxD identification. Specifically, diet measurement can be improved by incorporating (1) longitudinal diet measurements for more stable intake estimates and (2) objective metabolomic proxies for diet to complement self-reported data. The problem of small effect sizes can be improved by adapting the polygenic risk score (PRS) approach to create polygenic interaction scores (iPRS) that combine interaction effects across many variants. These strategies can be explored in the Trans-Omics for Precision Medicine (TOPMed) cohorts, which have self-reported diet at multiple timepoints and serum metabolomics. The MedDiet is a well-suited exposure with which to explore these approaches: it is commonly consumed, impacts metabolic disease risk, has evidence of GxD interactions, and has a validated metabolomic signature. The hypothesis to be explored is that improved dietary measurement approaches and iPRS will allow replication of known and discovery of novel GxDs. The principal investigator, Dr. Kenneth Westerman, Ph.D., is uniquely positioned to complete this work as an early- career scientist with a nutrition background, expertise in GxD analysis, and prior experience with TOPMed dietary data. He requires additional training to address current skill gaps in longitudinal data analysis, metabolomics, PRS, and leadership, facilitating the completion of the following research aims. Aim 1: To improve the reproducibility of standard gene-diet interaction models for metabolic risk factors by combining longitudinal measurements and metabolomic dietary proxies. Aim 2: To improve upon standard PRS for interaction testing by developing a polygenic score method that aggregates over genetic interaction effects Aim 3: To discover novel genetic factors modifying the association between a Mediterranean-style diet and metabolic risk factors. The proposed research will validate improved dietary modeling methods for GxD studies and create genetic scores that predict diet response. The associated training will help Dr. Westerman achieve independence in the field of precision nutrition and explore these ideas further in dietary trials in future grants.