Computational prediction of molecular tandem mass spectra using deep learning algorithms

Identifier:  TFG:5718

Handle: https://hdl.handle.net/20.500.11797/TFG5718

Authors:  Pérez Ribera, María Isabel

Abstract:
Disease diagnosis and personalized medicine based on metabolomics using changes in metabolite concentrations, are attracting the attention of more and more researchers. Nevertheless, compound identification remains a problem in most metabolomics studies based on mass spectrometry (MS), as the percentage of known MS molecular spectra is very low. We have proposed different methodologies to obtain the best prediction of the tandem mass spectra of molecules, comparing different types of neural networks. Also, we have obtained a very good prediction ability, achieving better results than the best in silico tool for the prediction of MS/MS spectra up to date.