Title: Improved Bayesian Network inference using relaxed gene ordering

Authors: Dongxiao Zhu, Hua Li

Addresses: Department of Computer Science, University of New Orleans, New Orleans, LA 70148, USA; Research Institute for Children, Children's Hospital, New Orleans, LA 70118, USA. ' Bioinformatics Center, Stowers Institute for Medical Research, 1000 E. 50th St. Kansas City, MO 64110, USA

Abstract: Bayesian Networks (BNs) have become one of the most powerful means of reconstructing signalling pathways in silico. Excessive computational loads limit the applications of BNs to learn larger sized network structures. Recent bioinformatics research found that signalling pathways are likely hierarchically organised. Genes resident in hierarchical layers constitute biological constraint, which can be readily used by BN structural learning algorithms to substantially reduce the computational load. We propose a constrained BN structural learning algorithm that solves the NP-complete computational problem in a heuristic manner. We demonstrate the utility of our algorithm in constructing two important signalling pathways in S. cerevisiae.

Keywords: Bayesian networks; biological pathways; microarrays; machine learning; Bayesian network inference; relaxed gene ordering; bioinformatics; structural learning algorithms; signalling pathways.

DOI: 10.1504/IJDMB.2010.030966

International Journal of Data Mining and Bioinformatics, 2010 Vol.4 No.1, pp.44 - 59

Published online: 14 Jan 2010 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article