Computational non-targeted analysis strategies for exploring the exposome chemical space

The number of unique structures in the exposome chemical space is ever-expanding due to their increasing release into the environment. Here, the chemical space refers to a vast collection of all possible organic structures, of which the subspace of possible organic molecules with a maximum of 17 atoms already amounts to 116 billion known structures. As of now, only a small fraction of the exposome chemical space has been identified. Therefore, further developments in mapping the chemical space are needed to understand the coverage and limitations of the current measurement techniques. A frequently used approach for analyzing exposome-related samples is non-targeted analysis (NTA) with liquid chromatography (LC) coupled to high-resolution mass spectrometry (HRMS), for which often a reversed phase (RP) selectivity is used. These samples can contain thousands of known and unknown chemicals, coming from a variety of compound classes. The main reoccurring aspect that needs to be considered with algorithm development for such data is the unknown composition of the samples, ensuring as little compound type discrimination as possible.
This thesis dives into enhancing the understanding of chemical analyzability using specific measurement methods (i.e., mapping the RPLC subspace), tackles issues related to the data pre- and postprocessing for NTA samples (i.e., naïve bayes isotope detection, fragment detection through cumulative neutral losses, and retention index alignment), and improving upon algorithm accessibility and modularity for a full NTA identification workflow (i.e., development of the modular open-source and open-access jHRMS toolbox).