Supervised Learning for Gene Regulatory Network Based on Flexible Neural Tree Model

Abstract

Gene regulatory network (GRN) inference from gene expression data remains a big challenge in system biology. In this paper, flexible neural tree (FNT) model is proposed as a binary classifier for inference of gene regulatory network. A novel tree-based evolutionary algorithm and firefly algorithm (FA) are used to optimize the structure and parameters of FNT model, respectively. The two E.coli networks are used to test FNT model and the results reveal that FNT model performs better than state-of-the-art unsupervised and supervised learning methods.