A smarter way to uncover cannabinoids missing from libraries

Cannabinoids comprise a chemically diverse group of meroterpenoids whose extensive isomerism, variable side-chain length, and frequent oxidative or rearranged derivatives lead to strongly overlapping yet characteristic MS/MS fragmentation patterns. In untargeted LC-MS/MS datasets, this combination of structural diversity and spectral similarity complicates annotation, particularly when reference spectra are sparse or unavailable. Library-based approaches, therefore, recover only a limited fraction of the cannabinoid-related chemical space that is routinely observed in experimental data. In this work, we apply MassQL to encode established cannabinoid fragmentation chemistry into rule-based queries. The resulting compendium covers major cannabinoid subclasses, including neutral and acidic cannabinoids, varinic analogs (C3 side-chain cannabinoids), and structurally modified derivatives, using combinations of diagnostic fragment ions, neutral loss patterns, adducts, and fragment co-occurrence logic. Importantly, class-level retrieval does not depend on complete or unambiguous precursor m/z information and can be driven solely by MS/MS evidence. Application of this framework to a publicly available untargeted LC-MS/MS dataset demonstrates that rule-based querying can recover known cannabinoids while highlighting additional features that share consistent cannabinoid-like fragmentation patterns. These features include putative analogs, transformation products, and derivatized forms that are not represented in current spectral libraries. At the same time, certain known features, such as in-source dehydrated ions, may be under-recovered depending on query design, illustrating current methodological limitations. This study demonstrates the feasibility and interpretability of chemically informed, rule-based MS/MS querying for cannabinoid discovery. Rather than replacing spectral library matching, MassQL-based class-level retrieval provides complementary hypothesis-generating evidence capable of expanding detectable cannabinoid chemical space beyond currently available reference spectra. The results also highlight the importance of polarity-aware fragmentation curation for reliable query-driven metabolomics workflows. MassQL class-level matches should be viewed as chemically informed hypotheses that complement, rather than replace, spectral library identification, while providing a basis for future systematic validation and benchmarking.