Multivariate functional analysis of the genetic basis of cancer

描述(由申请人提供)：拟议研究的目标是开发多元统计方法，利用现有的GWA数据对功能相关的生物信息进行联合分析，并进行关联和相互作用。具体地说，我们建议开发联合分析SNP集合(例如一个基因)中的多个SNP的方法，以及联合分析具有潜在可忽略和不可忽略缺失数据的多变量次级表型的方法。我们进一步建议开发全基因组范围的基因-基因相互作用分析方法，其中基因间的SNP-SNP相关性组将被一起分析以检测相互作用。此外，我们建议将先验知识整合到全基因组的关联或交互分析中；我们通过先验知识对预测因子集合进行分组，并设计灵活的正则化回归方法来约束参数估计，提高效率。拟议的研究是由GWA研究中的机会和需求推动的，我们建议的许多工作可以通过公开可访问的不同癌症的GWA数据以成本效益的方式实施，以提高我们对癌症的遗传基础和疾病病因的理解。

DESCRIPTION (provided by applicant): The objective of the proposed research is to develop multivariate statistical methods for joint analyses of functionally related biological information n association and interaction using existing GWA data. Specifically, we propose to develop methods for joint analyzing multiple SNPs in a SNP set (e.g. a gene), and methods for jointly analyzing multivariate secondary phenotypes with potential ignorable and non-ignorable missing data. We further propose to develop methods for genome-wide gene-gene interaction analysis, in which groups of between- gene SNP-SNP correlations will be analyzed together to detect interactions. In addition, we propose to integrate a priori knowledge in our genome-wide association or interaction analyses; and we group sets of predictors by a priori knowledge and design flexible regularized regression approaches to constrain the parameter estimation and achieve efficiency. The proposed research is motivated by opportunities and needs in GWA studies, and much of our proposed work can be cost-efficiently implemented with publicly accessible GWA data for different cancers to improve our understanding of the genetic basis and disease etiology of cancer.