Building blocks for exposome characterization Representation, extraction, and interpretation of chemical information in LC-HRMS-based non-target analysis

Characterizing the chemical exposome via non-target analysis (NTA) using liquid chromatography–high-resolution mass spectrometry (LC-HRMS) offers broad chemical coverage, yet extracting meaningful insights from complex spectral data remains challenging.
This thesis investigates methodological building blocks for improving the representation, extraction, and interpretation of chemical information in LC-HRMS-based NTA. First, a meta-analysis of published NTA studies evaluates current practices and reveals limitations in chemical-space coverage, reporting, and reproducibility. Next, approaches for chemical representation and feature prioritization are explored. The integration of complementary molecular fingerprints improves quantitative structure–activity relationship modelling, while methods based directly on mass spectrometric information demonstrate that environmentally relevant properties can be predicted even without prior compound identification.
The thesis further addresses interpretation of tandem mass spectrometry data. Systematic evaluation of spectral similarity metrics shows how metric choice and thresholding influence the discrimination of structurally similar isobaric candidates. Finally, the Fragment Ambiguity Score is introduced to quantify the diagnostic value of individual MS2 fragments by considering their occurrence within and across candidate compounds, revealing substantial differences in fragment informativeness among spectral libraries.
Together, these studies demonstrate how information at different stages of the NTA workflow can be more effectively represented, extracted, and evaluated. The resulting approaches provide building blocks toward more transparent, reproducible, and informative characterization of the chemical exposome.