|
Activity Number:
|
208
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 7, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #306037 |
|
Title:
|
Bayesian Variable Selection in Clustering High Dimensional Data with Substructure
|
|
Author(s):
|
Michael D. Swartz*+ and Marina Vannucci
|
|
Companies:
|
M. D. Anderson Cancer Center and Texas A&M University
|
|
Address:
|
1155 Pressler Street, Unit 1340, Houston, TX, 77030-3721,
|
|
Keywords:
|
Bayesian variable selection ; Bayesian mixture models ; reversible jump Markov chain Monte Carlo ; microarray analyses
|
|
Abstract:
|
Biological experimenters have incorporated microarray data into designed experiments, where the goal of inference is to uncover genes that are differentially expressed across the experimental groups. Commonly, researchers perform multiple ANOVA analyses with a multiple comparison correction. We cast the problem into one of Bayesian variable selection in clustering high dimensional data with substructure. The experimental groups naturally define substructure in the data, and clustering looks for similarities and differences among the subgroups, while the variable selection searches for those genes that best differentiate the subgroups. We have modified the reversible jump MCMC algorithm of Tadesse et al. (JASA, 2005) to account for experimental design substructure while exploring the cluster and feature space. We successfully employed the new algorithm to analyze preliminary data sets.
|
