Modelos bioinformáticos para la predicción de compuestos biológicamente activos en cáncer de colon

Zusammenfassung

For drug discovery it is necessary to find new compounds with specific chemical properties that allow the treatment of diseases. In recent years, the approach used in this search has an important component in computer science, with the rapid rise of machine learning techniques. With the objectives set by Precision Medicine and the new challenges generated, it is necessary to establish robust, standardized and reproducible computational methodologies to achieve the proposed objectives. Currently, predictive models based on Machine Learning have become very important in the step prior to preclinical studies, drastically reducing costs and research times in the discovery of new drugs. Chemoinformatics models can predict different outcomes (activity, chemical property, reactivity) in single molecules or complex molecular systems (catalysed organic synthesis, reactions, nanoparticles, etc.). Specifically, chemoinformatics prediction of enantioselectivity in complex catalytic systems is an important goal in organic synthesis and chemical industry. The Bronsted acid-catalysed enantioselective α-amidoalkylation reaction is a useful procedure for the production of new chiral catalysts (tools) or drugs (products). Enantioselectivity is sensitive to many factors, from the nature of the substrate and catalyst to the experimental conditions (solvent, temperature, etc.). Therefore, computational tools capable of predicting the enantioselectivity of these reactions are a valuable tool for the rational design of new catalysts and chiral products by organic synthesis chemists worldwide. This Doctoral Thesis has been elaborated, a very markedly multidisciplinary investigation, in which the efforts of three branches of science such as medicine, chemistry and computer science are combined. With regard to computing, the basis of the Doctoral Thesis, artificial intelligence techniques will be used to advance by creating new models. The bioinformatics tools developed serve as fundamental support for medical advances in the understanding of a disease as complex and multifactorial as cancer and also in the chemical advancement of the identification of structures of compounds that make them behave actively against colon cancer. . All this joint effort culminates in the development of new machine learning models based on artificial intelligence for the fight against colon cancer and the implementation of the first R library for the calculation of molecular descriptors, as well as the first public web server for its calculation online, validation of the models and a desktop version of the software for local use.