Activity Number:
|
428
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, August 9, 2006 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Biometrics Section
|
Abstract - #305743 |
Title:
|
Bayesian State-Space Models for Predicting Temporal Gene Expression Profiles
|
Author(s):
|
Yulan Liang*+ and Arpad Kelemen
|
Companies:
|
University at Buffalo and Niagara University
|
Address:
|
249 Farber hall, Buffalo, NY, 14214,
|
Keywords:
|
Bayesian approaches ; state space models ; prediction ; temporal gene expression ; MCMC algorithms ; deviance information criteria
|
Abstract:
|
Predictions the genome dynamic behavior is a challenging problem in genomic research and estimations the temporal correlation is one key. However, the unevenly short time courses with sudden change make the predictions difficult. In this paper, we develop two types of Bayesian State Space (BSS) models: multiple univariate time varying BSS and multivariate BSS models for predicting the gene expression profiles associated with diseases.In the univariate time-varying BSS model, we treat both stochastic transition and observation equations time-variant. In the multivariate model, we include temporal correlation structures with various prior settings. The unseen time points are treated as hidden state variables and are estimated by MCMC algorithms. We compared two type models using Deviance Information Criteria and applied our models to multiple tissues datasets.
|