|
Activity Number:
|
488
|
|
Type:
|
Invited
|
|
Date/Time:
|
Thursday, August 10, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #305214 |
|
Title:
|
Bayesian Mixture Models and Application to High-Throughput Data
|
|
Author(s):
|
Kim-Anh Do*+
|
|
Companies:
|
M. D. Anderson Cancer Center
|
|
Address:
|
Department of Biostatistics and Applied Mathematics, Houston, TX, 77030-4095,
|
|
Keywords:
|
mixture models ; gene expression ; mass spectrometry ; Dirichlet process ; beta distribution ; cancer experiments
|
|
Abstract:
|
The scientific world recently witnessed an explosion in the development of comprehensive, high-throughput methods for molecular biology experimentation. Our talk will focus on the development of Bayesian nonparametric mixture models with applications to two main platforms: microarray gene expression and mass spectrometry (MS) proteomic profiles. First, model-based inference is proposed for differential gene expression using a variation of the traditional Dirichlet process (DP) mixture models for the distribution of gene intensities under different conditions. Second, the unknown distribution of mass/charge ratios in MS data can be represented as a density estimation problem via a mixture of betas. We will present simulation studies and cancer-related experiments, contrasting the intrinsic differences in the technology and posterior inference that can incorporate multiplicities automatically.
|
