A Hierarchical Bayesian Model for the Gene Mapping of a Complex Disease

• Activity Number: 15
• Type: Contributed
• Date/Time: Sunday, August 11, 2002 : 2:00 PM to 3:50 PM
• Sponsor: ENAR
• Author(s): Michael Swartz*+ and Marek Kimmel and Christopher Amos
• Affiliation(s): Rice University and Rice University and U. T. M. D. Anderson Cancer Center
• Keywords: Complex disease mapping; ; Baeysain Model Selection ; Hierarchical models

Abstract

Mapping the genes for a complex disease, such as diabetes or rheumatoid arthritis, involves finding multiple genetic loci that may contribute to onset of the disease. Pairwise testing of the loci leads to the problem of multiple testing. To avoid multiple tests, we can look at haplotypes, or linear sets of loci; but this results in a contingency table with sparse counts, especially when using marker loci with multiple alleles. Using case-parent triad data, we develop a hierarchical Bayesian model, using a log-linear likelihood to model the probability of disease given genotype. We extend the Bayesian model developed by Thomas, et al. [(1995) Genetic Epidemiology 12:455-466] by developing prior distributions on the allele main effects and haplotype effects that model the genetic dependencies present in the HLA region of Chromosome 6. We also add a hierarchical level to allow for locus and allele selection. Thus, we cast the problem of identifying genetic loci relevant to the disease into a problem of Stochastic Search Variable Selection [George, E., and McCulloch, R.E. (1993) JASA 88:881-889]. Research supported by NCI grant R25 CA57730, Robert Chamberlain, Ph.D, P.I.