Desarrollo de una herramienta para la reconstrucción de metabolitos a partir del espectro de masas de los metabolitos.

Identifier:  TFG:2131

Handle: https://hdl.handle.net/20.500.11797/TFG2131

Authors:  Ruiz Botella, Manuel

Abstract:
Currently, metabolomics is the section of systems biology that has the most problems when it comes to integrating into biological models. This is because there is no way to sequence a metabolite as it does in other omic sciences. In metabolomics, access to databases is used to compare the spectrum of a metabolite with the stored spectra. Thanks to artificial intelligence, computer science and systems biology can be integrated to solve this limitation. Currently there is a model of artificial intelligence that is able to reconstruct molecules and generate new molecules. Our work is based on using an artificial intelligence model that is capable of reconstructing and generating new metabolites. In addition, we have developed a convolutional neuronal network that integrates with this first model and that allows us to reconstruct and generate new metabolites from the mass spectrum of a metabolite. Our model obtains a 92.83% reconstruction score for the metabolites that are used to generate it, and 24.54% for the metabolites that it has never seen. Our model is able to determine a molecule from its mass spectrum without the need that nobody has previously associated the mass spectrum of the molecule to the molecule. Completely surpassing the current methodology in metabolomics to determine a metabolite of a biological sample.