Using Dynamic Bayesian Networks to analyze genetic data
Authors: Zhen Wang and Andrew Wong.
Abstract:
In System biology, many statistical approaches are used to analyze gene expression data and infer gene regulatory network. Dynamic Bayesian Network is a well-known method that has produced promising results when used to analyze time-series data. Here, we adopt a Dynamic Bayesian Network algorithm with Markov Chain Monte Carlo (DBmcmc) developed by Dirk Husmeier to investigate gene expression data. We evaluated the algorithm’s performance on a synthetic yeast time-series data and then applied it to the rat CNS development temporal data to reverse engineer the gene network. The resulting network is then validated against previous studies on the same data set as well as evidence from biological databases and literature. The interactions documented in biological literature shows that DBmcmc was able to correctly identify subnetworks of interacting genes that were involved in the same biological pathways.