Optimised spectral processing and lineshape analysis in 2-dimensional J-resolved NMR spectroscopy based metabolomics

NMR spectroscopy is a primary analytical approach of metabolomics. Although 1D 1H NMR spectroscopy is versatile, highly reproducible and widely used, analysis of complex biological samples yields congested spectra with many overlapping signals. This makes metabolite identification and quantification challenging. 1H J-resolved (JRES) experiments spreads this high signal density into a second dimension, simplifying the spectral analysis. This thesis analyses the approaches and suitability of JRES spectroscopy to analyse metabolomics data. Firstly, the robustness of the JRES experiment is investigated. Using spectral relative standard deviation, benchmarks of spectral robustness can be compared between disparate processing techniques, sample types and analytical platforms. JRES spectra were found to be suitable for metabolomics experiments. Secondly, the application of standard metabolomic analysis methods to JRES spectra was examined. Using principal component analysis, the classification accuracy of 1D 1H and JRES spectra were investigated using several data sets. Alongside, three scaling methods were also evaluated. It was found that 2D JRES spectra and the glog transformation could produce 100% classification accuracy. Finally, spectral deconvolution of 2D JRES spectra from line-shape fitting was investigated Here, the mathematical functions describing the JRES line-shape, under several different processing conditions, are derived and used to create a semi-automated metabolite identification and quantification algorithm. Furthermore, possible quantitation errors arising from using JRES spectra are investigated, evaluating effects such as the overlapping of dispersive tails of nearby signals. In conclusion, the JRES experiment is a suitable for use in the field of metabolomics.EThOS - Electronic Theses Online ServiceGBUnited Kingdo