Colloquium: Tests for high-dimensional linear regression with application to genetic pathway analysis

Speaker: Dr. Yang Liu, Wright State University
Host: Dr. Zheng Xu
Time: Meet-n-Greet, 2:30 p.m.; Talk, 3 p.m.

ABSTRACT:

Linear regression analysis has become an important tool for investigating the association between a group of covariates and a continuous response. With the rapid development of technology, the number of covariates in regression analysis has increased rapidly and sometimes exceeds the sample size. We introduce an approach that can test for the significance of a set of covariates whose dimension could be greater than the sample size. The method can be used to detect both non-sparse weak signals and sparse-but-stronger signals in regression analysis. The asymptotic distribution for the proposed statistic is established. Simulation studies and a real application to a genome-wide association study of high-density lipoproteins are conducted to examine the finite-sample performance of the proposed method.

SPEAKER BIO:

Dr. Yang Liu is currently an Assistant Professor of Statistics at Wright State University. He received his Ph.D. in Statistics from Bowling Green State University in 2015, and continued his research as a Postdoc at Fred Hutchinson Cancer Research Center from 2015 to 2018. His research mainly focuses on high dimensional data analysis, graphical models, and statistical genetics/genomics.