Gene regulatory network reconstruction using dynamic Bayesian networks
High-content technologies such as DNA microarrays can provide a system-scale overview of how genes interact with each other in a network context. Various mathematical methods and computational approaches have been proposed to reconstruct GRNs, including Boolean networks, information theory, differential equations and Bayesian networks. GRN reconstruction faces huge intrinsic challenges on both experimental and theoretical fronts, because the inputs and outputs of the molecular processes are unclear and the underlying principles are unknown or too complex. In this work, we focused on improving the accuracy and speed of GRN reconstruction with Dynamic Bayesian based method. A commonly used structure-learning algorithm is based on REVEAL (Reverse Engineering Algorithm). However, this method has some limitations when it is used for reconstructing GRNs. For instance, the two-stage temporal Bayes network ( 2TBN ) cannot be well recovered by application of REVEAL; it has low accuracy and speed for high dimensionality networks that has above a hundred nodes; and it even cannot accomplish the task of reconstructing a network with 400 nodes. We implemented an algorithm for DBN structure learning with Friedman's score function to replace REVEAL, and tested it on reconstruction of both synthetic networks and real yeast networks and compared it with REVEAL in the absence or presence of preprocessed network generated by Zou and Conzen's algorithm. The new score metric improved the precision and recall of GRN reconstruction. Networks of gene interactions were reconstructed using a Dynamic Bayesian Network (DBN) approach and were analyzed to identify the mechanism of chemical-induced reversible neurotoxicity through reconstruction of gene regulatory networks in earthworms with tools curating relevant genes from non-model organism's pathway to model organism pathway.