Bayesian network approach to understand regulation of biological processes in cyanobacteria

Bayesian networks have extensively been used in numerous fields including artificial intelligence, decision theory and control. Its ability to utilize noisy and missing data makes it a good candidate to study biological systems. In this paper we propose the use of Bayesian network approach to study cellular response of cyanobacteria. We discuss how to combine individual gene expressions, obtained from microarrays generated using different platforms, to get biological process level behaviors. Biological processes carry more information towards understanding overall cell behavior. We then discuss several approaches available for identifying the structure of a Bayesian network and derive corresponding system level regulatory network for cyanobacterium, Synechocystis sp. PCC 6803. We discuss a method to quantify the strengths of the associations between different processes. The resultant network is used to simulate some of the experimental conditions and the responses of the network under those conditions are inferred. We show that these inferences agree with the observations made in the original experiments. Finally, we discuss how these type of networks could be helpful in making decisions on controlling the cellular activities so that the desired behaviors are achieved.