Deep Learning Approach To Gene Network Inference

Abstract

Gene regulatory network(GRN) inference remains a challenging problem in the field of bioinformatics. GRN contain valuable information needed to get a deeper understanding of the regulatory network. This could lead to advances in disease treatment or help drug discovery. Our approach to solving the problem of GRN inference is to train multilayered perceptrons (MLPs) to recreate the dynamics of the biological function. With the trained models able to module the dynamics, we hope to extract the underlying relationships between the species through feature attribution algorithms. We apply our method to a regulatory network for cell apoptosis and a network regulating the T-cell response to a pathogen.