Public defence in Information and Computer Science, M.Sc. Eric Bach

Title of the doctoral thesis: Machine learning methods for structural elucidation in untargeted metabolomics
Doctor's hat

Opponent: Assistant Professor Justin van der Hooft, Wageningen University, The Netherlands
Custos: Professor Juho Rousu, Aalto University School of Science, Department of Computer Science

The public defence will be organised on campus.

The thesis is publicly displayed 10 days before the defence in the publication archive Aaltodoc of Aalto University.

Electronic thesis

Public defence announcement

Studying organisms and our environment by characterizing the small molecules involved in biological process is one of the main objectives of metabolomics. Other than in genomics and proteomics, so called metabolites (small molecules) provide a dynamic view on biological systems. In food chemistry, for example, the effect of different diets is studied by investigating the metabolites contained in human bio-fluids. Or, in environmental science the occurrence of pesticides and their transformation products in, for example, surface waters is of interest due to their potential adversarial effects on living organisms.

To draw any biological conclusions the *structural elucidation*, that is the determination of the molecular structure (atom connectivity), of the measured molecules is a crucial step. Liquid-chromatography tandem-mass spectrometry (LC-MS²) is a widely used platform for such analyses. This thesis presents machine learning based approaches for the structural elucidation in LC-MS² experiments. Thereby, two novel ideas are studied. The first one is to include knowledge about the molecule alternatives one could observe, when analysis a signal, into the learning problem. The second main contribution is an approach to make use of so called retention orders, that is the order in which two molecules leave the LC system. These orders depend on the properties of the analyzed molecules. A model for the prediction of retention orders is presented and a framework to integrate them into the structural elucidation workflow.

The approaches presented in this thesis greatly improve the structural elucidation performance and are hence relevant to metabolomics community. Especially the utilization of retention orders is of great importance as it allows to resolve molecular structures in greater detail.

Contact details of the doctoral student:  [email protected]

  • Published:
  • Updated:
URL copied!