Network-Based Auto-Probit Modeling for Protein Function Prediction

• Activity Number: 342
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Xiaoyu Jiang*+ and David Gold and Eric Kolaczyk
• Companies: Boehringer Ingelheim Pharmaceuticals, Inc. and State University of New York at Buffalo and Boston University
• Keywords: Auto-probit ; Bayesian hierarchical model ; Gene Ontology annotation uncertainty ; Markov Chain Monte Carlo (MCMC) ; Protein function prediction ; STRING

Abstract

Protein function prediction has become a canonical problem in computational biology. We develop a hierarchical Bayesian probit-based framework for modeling binary network-indexed processes, with a latent multivariate conditional autoregressive Gaussian process to predict protein functions defined as terms in the Gene Ontology database. A natural extension of this framework can be used to model and correct for the high percentage of false negative annotations in GO, a serious short-coming endemic to biological databases of this type. Method performance is evaluated and compared with standard algorithms on weighted yeast protein-protein association networks from the STRING database. Results show that our basic method is competitive with these other methods, and incorporating the uncertainty in negative labels among the training data can yield nontrivial improvements in predictive accuracy.