A Boolean network inference from time-series gene expression data using a statistical method

Accurate gene regulatory network inference from time-series data is a challenging problem in computational biology. Recently, various methods have been developed to infer a gene regulatory network from time series data. Most of which are not scalable, because they limited the regulatory genes by 1 to 3. To resolve the problem, we propose an efficient statistical method called chi-square test for inferring gene regulatory network. A significant threshold p-value is considered to test the independence between target gene and regulatory gene in the chi-square test. Moreover, the p-value is corrected by Bonferroni Step-down (Holm) to reduce the false positives rate. A Boolean network is employed to model gene regulatory network because it is simple to implement, and it can capture network dynamics. To evaluate the performance of our method, we conducted an extensive simulation based on artificial datasets and compare the performance of our propose method with Relevance network and Pearson correlation methods, the experimental result shows that the proposed method significantly outperforms them with respect to structural accuracies. We further tested our method in two biological networks: SOS response of E.coli regulatory network and C.elegans cell cycle network. The simulation results demonstrated that our proposed method achieves better structural performance than relevance network and Pearson correlation methods also in the real dataset.