题目：Conditionally Symmetric Multidimensional Gaussian Mixtures for Testing Composite Null Hypotheses in Genetic Association Studies

主讲人：德克萨斯大学安德森癌症中心 Ryan Sun助理教授

主持人：统计学院 刘耀午教授

时间：2022年5月24日（周二）上午9:00-10:00

地点：Zoom: 863 519 982，密码：378884

报告摘要：

Inference for composite null hypotheses has become increasingly popular in genetic association settings such as pleiotropy studies, mediation analysis, and other similar investigations. The challenge is to determine whether all null hypotheses, as opposed to at least one, in a set of individual tests should be rejected. An appealing approach for genome-wide composite null inference is to apply multivariate extensions of the highly popular two-group model and calculate local false discovery rates (lfdr) for each set of hypotheses. However in practice, such a strategy is challenged by difficult multivariate density estimation. Furthermore, lfdr results can contradict findings from conventional z-statistics, which is troubling for a field that ubiquitously utilizes summary statistics. We propose a class of conditionally symmetric multivariate Gaussian mixture models that offers flexible options for multivariate density estimation while also ensuring monotonicity of lfdr with respect to summary statistics. The models are further studied through simulation and application to three composite null settings in genetic analysis of inflammatory diseases.

主讲人简介：

Ryan Sun is Assistant Professor in the Department of Biostatistics at the University of Texas MD Anderson Cancer Center. He completed his PhD and postdoctoral fellowship at Harvard University and joined MD Anderson in 2019. His research interests lie in developing novel statistical methodology that enables researchers to extract knowledge and insights from increasingly complex biomedical datasets. He also emphasizes applying these methods and disseminating the ideas to the broader biomedical research community. Currently he works heavily with genetics and genomics datasets to better understand the underlying causes of cancers and other diseases.