A Novel Chemical-Space-Dependent Strategy for Compound Selection in Non-target LC-HRMS Method Development Using Physicochemical and Structural Data

The virtual chemical space of substances, including emerging contaminants relevant to the environment and exposome, is rapidly expanding. Non-targeted analysis (NTA) by liquid chromatography-high-resolution mass spectrometry (LC-HRMS) is useful in measuring broad chemical space regions. Internal standards are typically used to optimize the selectivity and sensitivity of NTA LC-HRMS methods, assuming a linear relationship between structure and behavior across all analytes. However, this assumption fails for large, heterogeneous chemical spaces, narrowing measurable coverage to structurally similar compounds. We present a data-driven strategy for unbiased sampling of candidate structures for NTA LC-HRMS method development from extensive chemical spaces, such as the U.S. EPA's CompTox (>1 million chemicals). The workflow maximizes physicochemical/structural diversity using precomputed PubChem descriptors (e.g., molecular weight, XLogP) and grants LC-HRMS compatibility thanks to predicted mobility and ionization efficiency from molecular fingerprints. The resulting measurable compound lists (MCLs) provide broad, heterogeneous coverage for NTA method development, validation, and boundary assessment. Applied to the CompTox space, the approach yielded MCLs with greater chemical coverage and broader predicted LC-HRMS applicability than conventional "watch list" contaminants, offering a robust framework for enhancing NTA's measurable chemical space while preserving diversity.