Redes neurais recorrentes para inferência de redes de interação gênica utilizando cadeias de Markov

Abstract

Array technologies have made it strainghtforward to simultaneously monitor the expression pattern of thousands of genes. Thus, a lot fot data is being generated and the challenge now is to discover how to extract useful information from these data sets. Microarray data is highly specialized. It involves several variables in a nonlinear and temporal way, demanding nonlinear recurrent free models, which are complex to formulate and to analyse. So, this work proposes the use of Rucurrent Neural Networks(RNN) for data modeling, due to their learning hability of nonlinear and complex systems. Once a model is obtained with a RNN for the data, it is possible to extract rules to represent the knowledge acquired by them. From rule analisys, this work proposes the representation of the knowledge by Markov Chains model, which is easily visualized in the form of a graph of states, which show the interactions among the gene expression levels and their changes in time. In this work, we propose a new approach to microarra