A novel mutual information-based Boolean network inference method from time-series gene expression data

• DOI: 10.1371/journal.pone.0171097
• Publication Date: 2017-02-08T13:37:35Z
• Authors: Shohag Barman; Yung-Keun Kwon
• Keywords: Gene regulatory network; Boolean network; Network Dynamics

Background Inferring a gene regulatory network from time-series expression data in systems biology is challenging problem. Many methods have been suggested, most of which scalability limitation due to the combinatorial cost searching set genes. In addition, they focused on accurate inference structure only. Therefore, there pressing need develop method search genes efficiently and predict dynamics accurately. Results this study, we employed Boolean model with restricted update rule scheme capture coarse-grained dynamics, propose novel mutual information-based (MIBNI) method. Given as an input, first identifies initial using feature selection, then improves prediction accuracy by iteratively swapping pair between sets selected other Through extensive simulations artificial datasets, MIBNI showed consistently better performance than six well-known existing methods, REVEAL, Best-Fit, RelNet, CST, CLR, BIBN terms both structural accuracy. We further tested proposed two real datasets for Escherichia coli fission yeast cell cycle network, also observed results compared methods. Conclusions Taken together, promising tool predicting network.