A Machine Learning Approach for the Identification of a Treatment against Chagas Disease

Abstract

In this final degree project we have presented a machine learning approach to predict the biological activity of FDA approved drugs against T. cruzi. We believe that the proposed methodology will expand the state-of-art of machine learning in the Chagas disease drug discovery pipeline. We have obtained similar performance results with the work presented in but applied only to FDA approved drugs as a repurposing strategy. A final contribution of this work is the biological evaluation provided by the metabolic pathway analysis. This evaluation allows us to map FDA approved drugs onto T. cruzi metabolic pathways. This validation is useful because it incorporates important informa tion of how the drugs target T. cruzi. Finding a subset of drugs that come up from differently motivated experiments is promising. The fact that among our results are drugs that already have been tested in the past against Chagas disease is encouraging evidence that our approaches are able to produce reasonable candidates for drug repurposing. Additionally, the majority of the drugs present in our results were never tested against T. cruzi, confirming the novelty of our approaches.