Gene-Environment Interaction Analysis for Repeated Measures Data with AMMI Models

• Activity Number: 145
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Yi-An Ko*+ and Bhramar Mukherjee
• Companies: University of Michigan and University of Michigan
• Keywords: gene-environment interaction ; longitudinal data ; mixed model ; singular value decomposition ; Wishart distribution

Abstract

While most literature has focused on gene-environment interaction (G x E) in case-control studies, little attention has been given to sparse and efficient modeling of G x E in longitudinal studies with repeated measures data on quantitative traits. In a two-way table with categorical variables of G and E, a conventional saturated interaction model involves estimation for each G x E configuration. However, the degrees of freedom (df) for interaction test can grow quickly with increasing categories of G or E, which often results in loss of power for detecting an interaction. Additive main effects and multiplicative interaction (AMMI) entails a singular value decomposition of the interaction matrix and has been shown to have robust performance across a spectrum of interaction structures. Moreover, the df of G x E test can be reduced by a low rank approximation. We propose to use AMMI models and develop the test for G x E in longitudinal cohort studies. We apply the AMMI model to examining genetic modifying effects on the association between cumulative lead exposure and pulse pressure using data obtained from the Normative Aging Study.