Improving quantification and identification in metabolomics

Chapter 1 introduces and discusses metabolomics and related analytical techniques such as NMR, direct infusion-MS, GC-MS, and LC-MS. Next LC-MS-based untargeted metabolomics, including metabolite sample collection and preparation and untargeted LC-MS analysis platforms, are discussed. Chapter 2 apply the dynamic mass error theory to improve the quantification of metabolites. The aim of this project is the implementation of the dynamic binning algorithm, which considers the exact width broadening of a peak in mass spectra to construct extracted ion chromatograms (EICs), and which provides more accurate quantification than using EIC constructed with linear or constant thresholds implemented in the most widely used metabolomics software, XCMS. Chapter 3 describes applying dynamic thresholds to improve untargeted metabolite identification and includes the PEP score and q-value calculation to provide an estimated ensemble and individual error for untargeted metabolomics identification. Chapter 4 includes lipid identification and quantification. The transcriptomic data show that Vitamin B12, cofactor biosynthesis pathways, and ribosomal proteins are activated upon CP treatment. Especially the genes related to phospholipid biosynthesis and fatty acid are differentially expressed. These changes upon CP were further affirmed by the cell membrane’s composition change of fatty acids and phospholipids, as shown by both phospholipid composition and the membrane fluidity measurements. The growth inhibition of F. prausnitzii upon CP exposure can be explained by the cell membrane lipid composition and fluidity changes. Chapter 5 describes an R package with a graphic user interface, which enables untargeted metabolomics data pre-processing and analysis without programming skills.

Dissertation: https://hdl.handle.net/11370/a66df530-d039-4f3a-825c-7a552a652ca7