Functional analysis for LC-MS data in untargeted metabolomics : application to the Metnapar project

Metabolomics is a field whose aim is to identify small molecules involved in metabolic reactions to a drug, a disease or an environmental modification. In this field, the untargeted approach is the qualitative analysis of the largest set of metabolites possible. This approach generates a huge amount of data which requires biostatistics and bioinformatics tools to be treated. This master thesis is part of a group of theses whose goal is to evaluate and improve a data processing workflow for LC-MS untargeted metabolomics with an application to the Metnapar project. The focus of this thesis is on the steps of metabolites identifications and metabolic pathway interpretation. These two steps can be performed simultaneously using the web platform MetaboAnalyst. The mummichog algorithm used to this end is documented for a better understanding and evaluated in this thesis. Moreover, a tool of visualisation and reporting of the results is proposed in this thesis to present the results in an easily usable format for interpretation. Finally, the application of the complete workflow was performed on the data from two experiments of the Metnapar project. This master thesis has highlighted that the results of the complete workflow depends highly on the quality of the data. Furthermore, a change in the mummichog algorithm in MetaboAnalyst implementation seem to be problematic during the correction for multi-testing.