Time：2020/10/16 10:00

Location：Mingde Main Building 1016

Topic：Integrated Quantile Rank Test for gene-level associations in sequencing studies

Abstract:

Testing gene-based associations is the fundamental approach to identify genetic associations in sequencing studies. The best-known approaches include Burden and Sequence Kernel Association Tests (SKAT).  The gene-traits associations are often complex due to population heterogeneity, gene-environmental interactions, and various other reasons. The mean-based tests, including Burden and SKAT, may miss or underestimate some high-order associations that could be scientifically interesting. We propose a new family of gene-level association tests, which integrate quantile rank score processes while combining multiple weighting schemes to accommodate complex associations. The resulting test statistics have multiple advantages.  They are as efficient as the mean-based SKAT and Burden test when the associations are homogeneous across quantile levels and have improved efficiency for complex and heterogeneous associations. The test statistics are distribution-free, and could hence accommodate a wide range of distributions. They are also computationally feasible. We established the asymptotic properties of the proposed tests under the null and alternative hypothesis and conducted large scale simulation studies to investigate its finite sample performance.

Resume:

Dr. Tianying Wang is an assistant professor at the Center for Statistical Science in Tsinghua University. She earned her PhD from the Department of Statistics at Texas A&M University in 2018. Methodologically, she works on quantile regression, measurement error analysis, misspecified models, gene-environment interaction analysis, multivariate analysis and high-dimensional data analysis such as variable selection and classification. With respect to specific areas, she is primarily interested in cancer genomics, case-control studies, and epidemiology studies. She has worked on a variety of applied problems such as misspecified model subject to measurement error, high-dimensional binary classification with dimension reduction, semiparametric analysis of complex gene-environment interactions in case-control study.