Increased confidence of metabolite identification in high-resolution mass spectra using prior biological and chemical knowledge-based approaches

Mass spectrometry-based metabolomics aims to study endogenous, low molecular weight metabolites and can be used to examine a variety of biological systems. To substantially increase the accuracy of metabolite identification and increase coverage of the metabolome detected by high-resolution (HR) mass spectrometry I developed, optimised and/or employed several analytical and bioinformatics methods. Biological samples contain thousands of metabolites that are related through specific substrate-product transformations. This prior biological knowledge together with a mass error surface, which represents the mass accuracy of peak differences within mass spectra, were employed to significantly reduce the false positive rate of metabolite identification. To maximise the sensitivity of the Thermo LTQ FT Ultra mass spectrometer, the existing direct-infusion SIM-stitching acquisition parameters (Southam et al., 2007) were reoptimised, yielding a ca. 3-fold increase in sensitivity. Finally, relative isotopic abundance measurements (RIA) using HR direct-infusion MS were characterised on the two most popular Fourier transform MS instruments (FT-ICR and Oribitrap) using the reoptimised SIM-stitching acquisition parameters. Several novel observations regarding RIA measurements were reported. Utilising these RIA characterisations within a putative metabolite identification pipeline increased the number of single true empirical formula assignments compared to using accurate mass alone. To conclude, analytical and bioinformatics methods developed in this thesis have successfully facilitated the putative identification of hundreds of metabolites in several metabolomics studies.