Comparison of reversible-jump Markov-chain-Monte-Carlo learning approach with other methods for missing enzyme identification

Authors:
Bo Geng;Xiaobo Zhou;Jinmin Zhu;Y. S. Hung;Stephen T. C. Wong
Affiliations:
Department of Electrical and Electronic Engineering, University of Hong Kong, Hong Kong and HCNR - Center for Bioinformatics, Harvard Medical School, 3rd floor, 1249 Boylston, Boston, MA 02215, US ...;HCNR - Center for Bioinformatics, Harvard Medical School, 3rd floor, 1249 Boylston, Boston, MA 02215, USA;HCNR - Center for Bioinformatics, Harvard Medical School, 3rd floor, 1249 Boylston, Boston, MA 02215, USA;Department of Electrical and Electronic Engineering, University of Hong Kong, Hong Kong;HCNR - Center for Bioinformatics, Harvard Medical School, 3rd floor, 1249 Boylston, Boston, MA 02215, USA and Functional and Molecular Imaging Center, Brigham & Women's Hospital and Harvard Medica ...
Venue:
Journal of Biomedical Informatics
Year:
2008

Citing 3
Cited 1

Construction of genomic networks using mutual-information clustering and reversible-jump Markov-chain-Monte-Carlo predictor design

Signal Processing - Special issue: Genomic signal processing
Filling gaps in a metabolic network using expression information

Bioinformatics
Robust Full Bayesian Learning for Radial Basis Networks

Neural Computation

Growing enzyme gene networks by integration of gene expression, motif sequence, and metabolic information

Pattern Recognition

Quantified Score

Hi-index	0.00

Visualization

Abstract

Computational identification of missing enzymes plays a significant role in accurate and complete reconstruction of metabolic network for both newly sequenced and well-studied organisms. For a metabolic reaction, given a set of candidate enzymes identified according to certain biological evidences, a powerful mathematical model is required to predict the actual enzyme(s) catalyzing the reactions. In this study, several plausible predictive methods are considered for the classification problem in missing enzyme identification, and comparisons are performed with an aim to identify a method with better performance than the Bayesian model used in previous work. In particular, a regression model consisting of a linear term and a nonlinear term is proposed to apply to the problem, in which the reversible jump Markov-chain-Monte-Carlo (MCMC) learning technique (developed in [Andrieu C, Freitas Nando de, Doucet A. Robust full Bayesian learning for radial basis networks 2001;13:2359-407.]) is adopted to estimate the model order and the parameters. We evaluated the models using known reactions in Escherichia coli, Mycobacterium tuberculosis, Vibrio cholerae and Caulobacter cresentus bacteria, as well as one eukaryotic organism, Saccharomyces Cerevisiae. Although support vector regression also exhibits comparable performance in this application, it was demonstrated that the proposed model achieves favorable prediction performance, particularly sensitivity, compared with the Bayesian method.