ISBN: 978-981-11-3671-9 DOI: 10.18178/wcse.2017.06.058
Infer Gene Regulatory Network Using the Novel Nonlinear Differential Equation System
Abstract— Gene regulatory network (GRN) is very complex and nonlinear dynamics system. In this paper,
we present a novel nonlinear ordinary differential equation (ODE) model based on complex-valued flexible
neural tree (ODECVFNT) to improve the accuracy of GRN inference. Complex-valued flexible neural tree
(CVFNT) model is proposed to model the nonlinear regulation function in an ODE model. The hybrid
evolutionary method based on structure-based evolutionary algorithm and cuckoo search (CS) is used to
evolve the structure and parameter of ODECVFNT. Benchmark datasets from Dialogue for Reverse
Engineering Assessments and Methods challenge are used to test our method. Results reveal that our
proposed method can infer more correctly gene regulatory network than the popular method LASSO and
real-valued flexible neural tree (RVFNT) model.
Index Terms— gene regulatory network, complex-valued, flexible neural tree model, ordinary differential equation, cuckoo search
Wei Zhang, Bin Yang
School of Information Science and Engineering, Zaozhuang University, CHINA
Cite: Wei Zhang, Bin Yang, "Infer Gene Regulatory Network Using the Novel Nonlinear Differential Equation System," Proceedings of 2017 the 7th International Workshop on Computer Science and Engineering, pp. 337-343, Beijing, 25-27 June, 2017.