Two-Phase Study: Application to Gene-Environmental Interaction

• Activity Number: 418
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Jaeil Ahn*+ and Bhramar Mukherjee
• Companies: University of Michigan and University of Michigan
• Keywords: gene-environment interaction ; two-phase ; case-control ; Bayesian ; semi-parametric ; multivariate categorical

Abstract

We address the problem of exploring gene-gene and gene-environment interaction under two-phase sampling designs. Two-phase sampling designs are often more efficient than standard case-control sampling design. The Phase I sample of a two-phase study contains inexpensive covariates on every subject included in the study base, whereas Phase II sample is a stratified sample based on the case-control status and some of the available covariates. Expensive covariates are measured only in the phase II sample. The likelihood under this study design is similar to a missing data likelihood. We consider the special case when the Phase I sample has environmental covariates and Phase II samples contain genotype data. We then build a semi-parametric Bayesian model that data adaptively incorporates gene-gene and gene-environment independence assumption, if tenable, and offers an integrated computational