Reconstructing time series GRN using a neuro-fuzzy system

As a reverse engineering field, reconstructing a Gene Regulatory Network (GRN) from time series gene data has been a challenging issue in bioinformatics. This paper proposes a novel engineering framework that infers and reconstructs a gene regulatory network in terms of regulatory accuracy. Different from other statistical methods, the proposed framework uses features that represent the characteristics of time series datasets and selects the appropriate features of the time series data by using a neuro-fuzzy system. The proposed framework for reconstruction is based on a Neuro Network with Weighted Fuzzy Membership Function (NEWFM), which not only simplifies fuzzy inference and regulation model complexity but also improves the regulatory accuracy of reconstructing the GRN without minimizing the dynamic regulatory cycle. Finally, the proposed framework is evaluated with experimental results that demonstrate higher regulatory accuracy than previous algorithms.