GeneRegulatory Networks Inference withRecurrent Neural Network Models
Carl Isaac Wunsch · 2005
Large-scale timeseries geneexpression data generated fromDNA microarray experiments provide usanew meanstoreveal fundamental cellular processes, investigate functions of genes,and understand theirrelations and interactions. Toinfer generegulatory networks fromthese data witheffective computational tools hasattracted intensive efforts fromartificial intelligence andmachine learning. Here, weusea recurrent neural network (RNN), trained withparticle swarm optimization (PSO), toinvestigate thebehaviors ofregulatory networks. Theexperimental results, onasynthetic datasetand areal dataset, showthattheproposed modelandalgorithm can effectively capture thedynamics ofthegeneexpression time series andarecapable ofrevealing regulatory interactions between genes. I INTRODUCTION