Bayesian Methods for Genome-Wide Interacting QTL Mapping

描述（由申请人提供）：基因组技术的最新进展为确定遗传变异与健康和疾病的关系提供了前所未有的机会。大多数复杂的人类疾病都受到多基因（QTL）和环境因素相互作用网络的影响。相互作用（基因-基因和基因-环境）和遗传机制（例如，基因组印记、X连锁效应、多效性）在复杂疾病的遗传控制中发挥重要作用。复杂疾病的理想分析是同时考虑多个基因组基因座、环境因素和可能的相互作用，而不是一次考虑一个（或几个）基因座。尽管最近的方法发展，全基因组相互作用QTL的分析仍然是一个挑战。 拟议研究的目标是开发新的贝叶斯方法和软件，用于同时识别多个基因，环境因素及其相互作用，并探索重要的遗传机制（例如，基因组印记、X连锁效应、多效性）。所提出的方法将广义线性模型和分层建模的所有优点结合到相互作用基因的全基因组分析中，使我们能够处理各种类型的表型，同时分析许多相关变量，并开发稳定灵活的算法和软件。我们的具体目标是：1）开发新的贝叶斯广义线性模型和算法，用于定位实验杂交和群体关联研究中的互作QTL; 2）开发新的贝叶斯广义线性模型和算法，用于同时检测a）互作QTL和基因组印记，B）常染色体和X染色体上的互作QTL，以及c）多个相关性状的互作QTL; 3）通过广泛的模拟研究评估所提出的方法，将所提出的方法应用于多个真实的数据集，并提出用于多个相互作用QTL分析的模型检验和比较的贝叶斯方法; 4）将所提出的新方法并入我们的R/qtlbim软件（www.qtlbim.org）并发布扩展的R/qtlbim供公众使用。在这项提案中，我们专注于人类疾病的近亲繁殖动物模型，因为它们仍然是理解人类疾病病理机制的有力方法。然而，所提出的方法也可以扩展到人类的关联研究。预计该项目将对复杂疾病的遗传学/基因组学领域产生重要影响。

DESCRIPTION (provided by applicant): Recent advances in genomic technologies have provided unparalleled opportunities for identifying the relationship of genetic variation to health and disease. Most complex human diseases are influenced by interacting networks of multiple genes (QTL) and environmental factors. Interactions (gene-gene and gene- environment) and genetic mechanisms (e.g., genomic imprinting, X-linked effects, pleiotropy) play an important role in the genetic control of complex diseases. The ideal analysis of complex diseases is to simultaneously consider multiple genomic loci, environmental factors, and possible interactions rather than one (or a few) locus at a time. Despite recent methodological developments, genome-wide analysis of interacting QTL remains a challenge. The objectives of the proposed research are to develop new Bayesian methods and software for simultaneously identifying multiple genes, environmental factors, and their interactions, and exploring important genetic mechanisms (e.g., genomic imprinting, X-linked effects, pleiotropy). The proposed approach incorporates all advantages of generalized linear models and hierarchical modeling into genome-wide analysis of interacting genes, allowing us to deal with various types of phenotypes, to simultaneously analyze many correlated variables, and to develop stable and flexible algorithms and software. The specific aims of our proposal are to 1) develop new Bayesian generalized linear models and algorithms for mapping interacting QTL in experimental crosses and population association studies; 2) develop new Bayesian generalized linear models and algorithms for simultaneously detecting a) interacting QTL and genomic imprinting, b) interacting QTL on autosomes and X chromosome, and c) interacting QTL for multiple correlated traits; 3) evaluate the proposed methods by extensive simulation studies, apply the proposed methods to multiple real data sets, and propose Bayesian methods of model checking and comparison for multiple interacting QTL analysis; and 4) incorporate the proposed new methods into our R/qtlbim software (www.qtlbim.org) and release the extended R/qtlbim for public use. In this proposal, we focus on inbred animal models of human diseases because they continue to be a powerful approach to understanding the pathological mechanisms of human diseases. However, the proposed methods can also be extended to association studies in humans. The project is expected to make an important impact on the field of genetics/genomics of complex diseases.