Identifying rare haplotype-environment interactions using Logistic Bayesian Lasso

描述(申请人提供)：罕见的变异被认为是在癌症等复杂疾病中发现\缺失遗传性的关键。这些变异现在可以使用下一代测序技术进行基因分型；然而，罕见的单倍型也可能是基因组范围关联研究(GWAS)提供的常见SNP组合的结果。在这方面，可能有大量的宝藏尚未从GWAS的数据中挖掘出来，以探索常见疾病罕见变异假说。最近，我们提出了一种名为Logistic Bayesian Lasso(LBL)的方法来在病例对照环境中识别与罕见单倍型的关联。最小二乘法是惩罚回归方法套索的贝叶斯方法的改编。我们的方法能够剔除不相关的(特别是常见的)单倍型，以实现足够的降噪，从而可以更容易地检测到关联的稀有单倍型中包含的信号。利用LBL，我们首次在补体因子H(CFH)基因中发现了年龄相关性黄斑变性(AMD)的一种特殊的罕见单倍型。除了罕见的变异外，基因-环境相互作用(GxE)被认为是丢失遗传性的另一个重要因素。LBL有一个灵活的框架，可以结合非遗传(环境)协变量和基因-环境相互作用。在这个项目中，我们提出了探索癌症流行病学中罕见单倍型与环境因素之间相互作用的方法，首先是在简单随机抽样的背景下，然后是分层随机抽样。我们将开发有或没有基因-环境独立性假设的方法。这些方法将通过在各种设置下的模拟来广泛研究。它们将被应用于从NIH的基因类型和表型数据库(DBGaP)和AMD数据中获得的几个癌症数据集。此外，分层抽样的方法将被用于分析NCI赞助的肾癌病例对照研究，其中对照是通过与病例进行频率匹配的分层抽样来选择的。我们将在一个文件齐全、用户友好的软件中实施建议的方法，并将其提供给更大的科学界。

DESCRIPTION (provided by applicant): Rare variants have been heralded as key to uncovering \missing heritability" in complex diseases such as cancers. These variants can now be genotyped using next-generation sequencing technologies; nonetheless, rare haplotypes may also result from combination of common SNPs available from Genome-Wide Association Studies (GWAS). In this regard, there may be a great deal of treasure that are yet to be mined from the GWAS data to explore the common disease rare variant hypothesis. Recently, we have proposed an approach named Logistic Bayesian LASSO (LBL) to identify association with rare haplotypes in a case-control setting. LBL is an adaptation of the Bayesian counterpart of penalized regression approach LASSO. Our approach is able to weed out unassociated (especially common) haplotypes to achieve enough noise reduction so that the signals contained in the associated rare haplotypes can be more easily detected. Using LBL, we were able to implicate a specific rare haplotype for Age-related Macular Degeneration (AMD) in the Complement Factor H (CFH) gene for the first time. In addition to rare variants, gene-environment interaction (GXE) is believed to be another important contributor to missing heritability. LBL has a flexible framework that can incorporate non-genetic (environmental) covariates and gene- environment interactions. In this project we propose methods for exploring interactions between rare haplotypes and environmental factors in cancer epidemiology, rst in the setting of simple random sampling and then for stratified random sampling. We will develop methods both with and without the assumption of gene-environment independence. The methods will be extensively studied through simulations under a variety of settings. They will be applied to several cancer datasets available from NIH's database of Genotypes and Phenotypes (dbGaP) and the AMD data. Further, the method for stratified sampling will be used to analyze the NCI-sponsored Kidney Cancer Case-Control Study, wherein the controls were selected by stratified sampling using frequency matching with cases. We will implement the proposed methods in a well-documented user-friendly software and make it available to the larger scientific community.