Statistical Genetics and Genomics for Epidemiologic Research

项目总结/摘要 全基因组关联研究已经确定了前所未有的数量与疾病相关的遗传变异 然而，遗传风险等位基因的分子机制和临床意义在很大程度上是未知的。流行病学 研究正在将其范围扩展到更多的转化和机制研究，将遗传风险变异与 环境暴露、干预、基因表达和表观遗传学。受基于人群的研究的启发， 前列腺癌的研究，我们将开发新的翻译主题的统计方法：如何搜索 预测个体和亚组干预效果的基因型？如何识别表观遗传改变 环境和基因组的界面如何评估可改变风险因素的因果中介效应， 与疾病结果相关的分子改变？这些主题提出了尚未解决的统计挑战，因为 高维度和复杂建模。有待开发的具体统计方法包括高维基因- 治疗交互作用，多位点区域关联，中介分析，工具变量分析，孟德尔 随机化、收缩和正则化。 该项目的方法学研究主要是由前列腺癌驱动的，前列腺癌是最常见的非皮肤 癌症是美国男性癌症死亡的第二大原因，在他的一生中占六分之一。这个项目 巢在高度完成的财团研究（PCPT/SELECT/PRACTICAL/PCPS），所有这些都产生了 对前列腺癌研究产生深远影响。该项目的独特之处在于， 将与正在进行的分析无缝集成，确保立即翻译。我们的跨学科研究 团队一直积极从事统计遗传学和基因组学，并进行分子流行病学研究。 PI在高维方法、分子生物标志物和遗传流行病学方面拥有丰富的专业知识。这 该项目将对癌症病因学、预防和治疗结果的方法产生深远的影响。

Project Summary/Abstract Genome-wide association studies have identified an unprecedented number of genetic variants associated with disease risk, yet molecular mechanisms and clinical implications of genetic risk alleles are largely unknown. Epidemiologic research is extending its reach to more translational and mechanistic studies, integrating genetic risk variants with environmental exposures, interventions, gene expression and epigenetics. Motivated by population-based studies for prostate cancer research, we will develop statistical methods for emerging translational topics: how to search for genotypes that predict individual and subgroup intervention effects? how to identify epigenetic alterations that may be an interface of the environment and the genome? how to assess causal mediation effect of a modifiable risk factor or a molecular alteration in relation to disease outcomes? These topics present unmet statistical challenges because of high dimensionality and complex modeling. Specific statistical methods to be developed include high-dimensional gene- treatment interaction, multi-locus regional association, mediation analyses, instrumental variable analyses, Mendelian randomization, and shrinkage and regularization. The methodological research in this project is driven primarily by prostate cancer, the most common noncutaneous cancer and the second leading cause of cancer death in American men, a ecting one in six in his lifetime. This project nests in highly-accomplished consortium studies (PCPT/SELECT/PRACTICAL/PCPS), all of which have generated far-reaching impact on prostate cancer research. The unique feature of this project is that methodological development will be seamlessly integrated with ongoing analyses, ensuring immediate translation. Our transdisciplinary research team has been actively engaged in statistical genetics and genomics, and conducting molecular epidemiological studies. The PI brings a wealth of expertise in high-dimensional methods, molecular biomarkers, and genetic epidemiology. This project will have a far-reaching impact on methodologies in cancer etiology, prevention, and treatment outcomes.