Genetic Association and Personalized Medicine

 描述(申请人提供)：尽管最近发表的全基因组关联研究(GWAS)定位了许多与疾病相关的遗传变异，但它们只解释了极小比例的可遗传表型变异，这表明由于遗传异质性(即与复杂性状相关的多个遗传变异)、普通遗传变异的小到中等效应大小以及当前分析方法有限的统计能力，只确定了一小部分原因位点。另一方面，Gwas数据也为个性化医学提供了一个令人兴奋的机会，旨在根据个人的临床和遗传信息将最合适的治疗或干预分配给他/她。然而，将GWAS数据转化为个性化医学实践还有相当长的路要走，这主要是由于缺乏强大的分析方法。这项研究致力于利用高维遗传和临床数据对个性化医学中的几个新兴主题进行研究。在前一次资助期间在惩罚回归和分类方面取得的进展的基础上，我们建议开发创新和强大的统计方法，用于GWAS数据，以发现新的基因途径并将其用于个性化医学。特别是，我们的目标是发现包含SNPs的从头基因途径，这些SNPs对复杂的疾病和性状具有单独较弱但总体上较强的影响。我们结合了现有的肺健康研究(LHS)临床数据和来自两个不同来源的Gwas数据，应用开发的统计方法来探索遗传变异和基线临床变量如何可能改变吸烟干预的效果，以及如何为任何给定的受试者确定最优的个体化干预规则。

 DESCRIPTION (provided by applicant): Although recently published genome-wide association studies (GWASs) have localized many disease-associated genetic variants, they only account for tiny proportions of heritable phenotypic variations, suggesting that only a small fraction of causal loci have been identified, due to genetic heterogeneity (i.e. multiple genetic variants associated with a complex trait), confirmed small to modest effect sizes of common genetic variants and limited statistical power of current analysis methods. On the other hand, GWAS data also offer an exciting opportunity for personalized medicine, aiming to assign the most suitable treatment or intervention to an individual based on his/her clinical and genetic information. However, there is still quite a distance in translating GWAS data to practice of personalized medicine, largely due to the paucity of powerful analysis methods. This research is devoted to several emerging topics in personalized medicine with high-dimensional genetic and clinical data. Building on the advances in penalized regression and classification made during the previous funding period, we propose developing innovative and powerful statistical methods for GWAS data to discover novel gene pathways and utilize them in personalized medicine. In particular, we aim to discover de novo gene pathways containing SNPs with individually weak, but collectively strong, effects on complex disease and traits. We combine the available Lung Health Study (LHS) clinical data and GWAS data from two different sources, applying the developed statistical methods to explore how genetic variants and baseline clinical variables possibly modify the effects of smoking interventions, and how to determine an optimal individualized intervention rule for any given subject.