Gene networks inference from expression data using a recurrent neuro-fuzzy approach

The reverse engineering paradigm is given increasing attention in computational molecular biology lately. One of the goals is to understand how gene regulatory networks (complex systems of genes, proteins and other molecules) function and interact to carry out specific cell functions. We present an approach for inferring the complex causal relationships among genes from microarray experimental data based on a recurrent neuro-fuzzy method. The method derives information on the gene interactions in a highly interpretable form (fuzzy rules) and takes into account dynamical aspects of genes regulation through its recurrent structure. We tested our approach on a set of genes known to be highly regulated during the yeast cell-cycle. The retrieved gene interactions correspond to the ones validated by previous biological studies, while our method surpasses previous computational techniques that attempted gene networks reconstruction, being able to retrieve significantly more biologically valid relationships among genes. At the same time, our method is able to predict time series for the expression of the genes based on the information extracted from a training subset of the data. The results prove highly accurate prediction capability.