General Framework for Rare Variant Analysis: Gene-Environment Interaction and Multiple Trait Analysis in Family Samples

• Activity Number: 84
• Type: Contributed
• Date/Time: Sunday, August 3, 2014 : 4:00 PM to 5:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Wei Gao*+ and George T. O'Connor and Josée Dupuis
• Companies: and Boston University School of Medicine and Boston University School of Public Health
• Keywords: rare variant analysis ; multiple trait analysis ; repeated measures analysis ; gene-environment interaction ; family samples ; linear mixed effects model

Abstract

There are many statistical methods for rare variant association analysis. The sequence kernel association test (SKAT) has been shown to be powerful in various scenarios. The family-based SKAT (famSKAT) is an extension of SKAT to account for familial correlation. We developed a general SKAT framework that is applicable to family sample, repeated phenotype measurements and multiple traits. The general SKAT (genSKAT) takes correlations between multiple measurements into consideration in addition to accounting for family structures. SKAT and famSKAT have inflated type I error if correlations between repeated phenotype measurements or multiple traits are inappropriately ignored. In contrast, genSKAT has the correct type I error. We further discussed gene-environment interaction tests under the genSKAT framework. In some studies, each individual has more than one measurement. For complex diseases, the environment exposure and its interaction with the genes may affect more than one trait. The gene effects may also vary over time. The genSKAT takes advantage of multiple traits/measurements and possible gene-environment interactions, and improves power to detect genetic associations.