Joint analyses of multiple genetic variants and multiple trait phenotypes

这一研究计划的长期目标是开发统计方法和计算工具，以解决复杂人类遗传研究中出现的问题。最近的高通量技术提供了以低成本收集海量数据的有效手段。然而，基因发现的一个主要障碍是开发强大而稳健的统计方法，充分利用这些数据提供的广度和深度。拟议的研究重点是开发新的多变量方法，用于联合分析多个遗传变异和多个性状表型，这些方法对模型错误规范具有鲁棒性。我们评估了在统一框架下分析一组罕见(或常见)遗传变异的方法，并提供了统计见解来回答一些反复出现的问题，并澄清了人类遗传学和统计遗传学社区中存在的一些混淆。这进而指导了稳健关联检验统计量的发展，这些统计量涉及总体分层、测量误差和非随机抽样设计。我们还讨论了在遗传多效性研究背景下联合分析多个相关表型的概念和统计学挑战，并提供了一些可行的解决方案。所提出的方法将为新兴的基因研究提供各种稳健和强大的多变量统计，促进我们对下一代基因数据的解释和理解。

The long-term objective of this research program is to develop statistical methods and computational tools to address issues arising from genetic studies of complex human traits.The recent high-throughput technologies provide efficient means to collect vast amount of data with low cost. However, a major hurdle to gene discoveries is the development of powerful and robust statistical methods that take full advantage of the breadth and depth offered by such data.The proposed research focus on developing novel multivariate methods for joint analyses of multiple genetic variants and multiple trait phenotypes that are robust to model misspecifications. We evaluate methods for analyzing a group of rare (or common) genetic variants under a unified framework and provide statistical insights to answer some of the recurring questions and clarify some of the confusions that exist in the communities of human genetics and statistical genetics. This in turn guides the development of robust association test statistics that deal with, population stratification, measurement errors and non-random sampling design. We also discuss the conceptual and statistical challenges in joint analyzing multiple correlated phenotypes in the context of genetic pleiotropy studies, and provide some viable solutions to the problem. The proposed methodology will provide emerging genetic studies with various robust and powerful multivariate statistics, advancing our interpretation and understanding of next generation genetic data.