Mass spectrometry coupled to liquid chromatography (LC-MS) is routinely used for metabolomics studies. While steps in data acquisition are fairly standardised and automated, structural metabolite identification still depends on manual curation and expert knowledge, forming a major bottleneck in LC-MS based pipelines. The work presented in this thesis represents a novel data processing strategy, which aids metabolite identification through
deliberate us of the the correlation structure that exists between spectral features, as well as chromatographic profile and data acquisition order. This strategy aligns features originating from the same chemical entity across all samples as a group, ensuring that chemically-related features are accurately aligned despite fluctuations in the chromatographic and mass spectrometric measurements occurring during the experimental run time. Spectral features aligned in this way are consequently matched to in-house chemical standards databases more efficiently and accurately, on account of the retained and chemically-relevant spectral information. This pipeline has been developed and is presented as an open-source R package - massFlowR. This thesis demonstrates the utility of massFlowR with simulated data, as well as an open-source urine metabolomics study DEVSET, and a large-scale cohort study AIRWAVE, where the performance of massFlowR is compared with the widely-used package XCMS.