A biocomputational application for the automated construction of large-scale metabolic models from transcriptomic data

Abstract

Cancer is a very complex disease with particular metabolic features that turn it into a very difficult system to approach from the solely experimental research. Therefore a systems biology approach is absolutely required to shed light on the subjacent mechanisms in order to derive reliable predictions about cancer evolution or behavior. In the present work, we developed a computational application that implements existing methods using a general metabolic model and gene expression data to generate cancer-specific models. As a working example, we used expression data of breast cancer cell lines to generate 3 models where we could consistently observe cancer-specific alterations at aldehyde dehydrogenase in the glycolysis, which is related to breast cancer stem cells and also in a reaction of glutathione peroxidase related to cancer chemoresistance. This computational application of metabolic modeling can be extended easily to add more methods of model generation and can be adapted to automatically construct personalized metabolic models that could be helpful in the prediction of chemotherapy response and find personalized cancer targets to optimize cell death and overcome therapy resistance.