|
Activity Number:
|
269
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #305864 |
|
Title:
|
Bayesian Hidden Markov Modeling of Array CGH Data
|
|
Author(s):
|
Subharup Guha*+
|
|
Companies:
|
Harvard School of Public Health
|
|
Address:
|
199 Park Drive, Boston, MA, 02215,
|
|
Keywords:
|
genomic alterations ; cancer ; DNA ; MCMC
|
|
Abstract:
|
Genomic alterations have been linked to the development of cancer. The technique of Comparative Genomic Hybridization (CGH) yields data that provide information about changes in DNA copy number. As increasing amounts of array CGH data become available, there is a growing need for automated algorithms that detect copy number gains and losses based on statistical considerations. We take a Bayesian approach, relying on the hidden Markov model to account for the dependence in the data. Localized amplifications (associated with oncogenes) and deletions (associated with tumor suppressors) are identified using posterior probabilities based on an MCMC sample. Publicly available data on pancreatic adenocarcinoma are analyzed. Comparisons are made with widely used methods to illustrate the reliability and success of the technique.
|
