|
Activity Number:
|
622
- Statistical Methods for Genome- and Epigenome-Wide Association Studies and Gene Environment Interactions
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 1, 2019 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Genomics and Genetics
|
|
Abstract #305353
|
|
|
Title:
|
Structured Variable Selection for High-Dimensional Data, with Applications in Gene-Environment Interactions in Longitudinal Studies
|
|
Author(s):
|
Fei Zhou* and Jie Ren and Cen Wu
|
|
Companies:
|
Kansas State University and Kansas State University and Kansas State University
|
|
Keywords:
|
Gene-environment interaction;
structured variable selection;
high-dimensional data;
longitudinal study;
penalization
|
|
Abstract:
|
Gene-environment interactions play an important role in the occurrence of complex diseases beyond genetic and environmental main effects. Investigations of gene-environment interactions help us better understand the genetic basis of disease phenotype. Multiple studies have been conducted to identify important gene-environment interactions by using variable selection methods. However, in longitudinal studies, penalization methods have not been widely developed to accommodate the structured interactions. We proposed a penalized variable selection method that identifies important gene-environment interactions as well as genetic main effects and respects the strong hierarchical structure for high-dimensional genomics data with longitudinal responses. Simulation studies have shown the proposed method outperforms several other benchmark methods, with superiority in terms of both identification accuracy and prediction performance. Analysis on high dimensional data from longitudinal studies shows that the proposed method identifies markers of important implications and leads to satisfactory predictions.
|
Authors who are presenting talks have a * after their name.
