Analysis of DNA Repair Studies through Bayesian Hierarchical Models for Mixtures

• Activity Number: 277
• Type: Contributed
• Date/Time: Tuesday, August 9, 2005 : 10:30 AM to 12:20 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Abel Rodriguez*+ and Dunson David
• Companies: Duke University and National Institute of Environmental Health Sciences
• Keywords: Comet assay ; Factor analysis ; Finite mixture model ; Molecular epidemiology ; Stochastic search ; Hierarchical model

Abstract

Single-cell electrophoresis, also known as the comet assay, can be used to study DNA damage and repair mechanisms by measuring the frequency of DNA strand breaks for individual cells in samples collected at repeated times before and after exposure to a genotoxic agent. In such studies, the distribution of the measured surrogate of DNA damage across cells in a sample can vary substantially for different cell lines and followup times. Because the distributional shapes differ, standard hierarchical models are not adequate. We develop a Bayesian approach based on a finite mixture of normals, which allows the mixture weights to shift dynamically across time and cell lines. Specifically, the weights are assigned a hierarchical model that includes a factor structure, with cell line-specific latent traits measuring baseline damage, susceptibility to induced damage, and rate of DNA repair. A Gibbs sampler is developed for posterior computation and methods for inferences on the covariance structure. The proposed approach can be used to assess heterogeneity among individuals and for identifying the responsible genotypes.