Automated discovery of conjugates drives a reverse-metabolomics strategy: From LC-HRMS-based phase II prediction to GC-HRMS validation of phase I metabolites of 6β-chlorotestosterone.

Anti-doping and forensic science detection rely fundamentally on target-based methodologies, especially for metabolite screening and identification. Traditional manual screening is constrained by low efficiency and knowledge bias. Conversely, although current metabolomics and machine-learning approaches efficiently calculate lots of potential biomarkers using statistical methods, these findings necessitate further rigorous screening to isolate true biomarkers that satisfy the requirements of both anti-doping and forensic science. To bridge the gap between machine learning and manual screening, we developed an innovative reverse-metabolomics strategy: utilizing LCHRMS for Phase II conjugate automated prediction as well as GCHRMS for targeted Phase I tentative validation. In the study of 6β‑chloro-testosterone (6β-Cl-T) metabolism, while LCHRMS (via Compound Discoverer software) initially predicted 244 putative phase II metabolites, the integration of GCHRMS, following the release of Phase I metabolites and subsequent derivatization, acted as a high-precision filter to further elucidate these candidates. 74 of the 244 candidates were putatively validated as candidate biomarkers. Eliminating high-throughput false positives, this workflow boosts data confidence in anti-doping and clinical metabolomics.