Online Program Home
My Program

Abstract Details

Activity Number: 412 - Data Science and Machine Learning Topics
Type: Contributed
Date/Time: Tuesday, July 30, 2019 : 2:00 PM to 3:50 PM
Sponsor: Section on Statistical Computing
Abstract #306812 Presentation
Title: A General Multivariate Linear Mixed Model for Detecting Gene by Environment Interactions
Author(s): Hyeonju Kim* and Saunak Sen
Companies: University of Tennessee Health Sci Ctr and University of Tennessee Health Sci Ctr
Keywords: multivariate linear mixed models; gene by environment (GxE) interactions; genetic similarity; QTL

We develop a multivariate linear mixed model for a general framework to detect gene by environment (GxE) interactions. Our model incorporates low- and high-dimensional phenotype covariates to capture the correlation between phenotypes across multiple environments. Phenotypes are assumed to be correlated due to genetic similarity (measured by genome-wide markers) and environmental similarity (measured by weather information). The correlation structure is motivated via a multivariate linear model with small random GxE effects. The model is estimated using an expectation conditional maximization (ECM) embedded in an accelerated gradient descent scheme. We illustrate the methods, accuracy, and performance time through quantitative trait loci (QTL) analysis of Arabidopsis data with annotated weather information in 3 years by 2 locations.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program