Information management and multivariate analysis techniques for metabolomics data

Among the so-called "omics" disciplines,metabolomics has been receiving considerable attention over the last few years. Metabolomics is the large-scale study ofmetabolites that are smallmolecules within cells, biofluids and tissues, produced as a result ofmetabolism. The growing interest inmetabolomics has been encouraged by rapid advances inmetabolic profiling techniques and by technological developments of the diverse analytical platforms, including proton NucleicMagnetic Resonance (1H NMR), Gas Chromatography-Mass Spectrometry (GC-MS) and Liquid Chromatography-Mass Spectrometry (LC-MS), used for extracting metabolic profiles. The output generated from these experimental techniques results in the production of a huge amount of data and information. This thesis attempts to provide an overview of the analytical technologies, the resources and databases employed in this emerging discipline, and ismainly focused on the following two aspects: (i) the challenges of handling the large amounts of data generated and managing the complex experimental processes needed to produce them; (ii) the techniques for the multivariate analysis of metabolomics data, with a special emphasis on methods based on the randomforest algorithm. To this aim, a detailed description and explanation of QTREDS, a software platform designed for managing, monitoring and tracking the experimental processes and activites of "omics" laboratories is provided. In addition, a thorough elucidation of the software package RFmarkerDetector, available through the Comprehensive R Archive Network (CRAN), and a description of the multivariate analysis techniques it implements, is also given.