|
Activity Number:
|
235
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 5, 2008 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #301460 |
|
Title:
|
Network-Based Auto-Probit Modeling for Protein Function Prediction
|
|
Author(s):
|
Xiaoyu Jiang*+ and Eric Kolaczyk+ and David Gold
|
|
Companies:
|
Boston University and Boston University and Boston University
|
|
Address:
|
111 Cummington Street, Boston, MA, 02215, Department of Mathematics and Statistics, Boston, MA, ,
|
|
Keywords:
|
Protein function prediction ; Conditional autoregressive ; Network
|
|
Abstract:
|
Predicting the functional roles of proteins based on various genome-wide data, such as protein-protein interaction networks (PPI), has become a canonical problem in high-throughput computational biology. Approaching the problem as binary classification, we propose a network-based extension of the spatial auto-probit model. In particular, we develop a fully Bayesian probit-based framework, with a latent multivariate conditional autoregressive (CAR) Gaussian process, where the latter encodes a measure of protein functional similarity influenced by network topology. We use this framework to predict protein functions, for function defined in the Gene Ontology (GO) database, a popular rigorous vocabulary for biological functionality. Markov Chain Monte Carlo methods are used to gain posterior estimates. A cross-validation study is performed on the data from the yeast Saccharomyces cerevisiae.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2008 program |
