Nontarget high-resolution mass spectrometry screening (NTS HRMS/MS) can detect thousands of organic substances in environmental samples. However, new strategies are needed to focus time-intensive identification efforts on features with the highest potential to cause adverse effects instead of the most abundant ones. To address this challenge, we developed MLinvitroTox, a machine learning framework that uses molecular fingerprints derived from fragmentation spectra (MS2) for a rapid classification of thousands of unidentified HRMS/MS features as toxic/nontoxic based on nearly 400 target-specific and over 100 cytotoxic endpoints from ToxCast/Tox21. Model development results demonstrated that using customized molecular fingerprints and models, over a quarter of toxic endpoints and the majority of the associated mechanistic targets could be accurately predicted with sensitivities exceeding 0.95. Notably, SIRIUS molecular fingerprints and xboost (Extreme Gradient Boosting) models with SMOTE (Synthetic Minority Oversampling Technique) for handling data imbalance were a universally successful and robust modeling configuration. Validation of MLinvitroTox on MassBank spectra showed that toxicity could be predicted from molecular fingerprints derived from MS2 with an average balanced accuracy of 0.75. By applying MLinvitroTox to environmental HRMS/MS data, we confirmed the experimental results obtained with target analysis and narrowed the analytical focus from tens of thousands of detected signals to 783 features linked to potential toxicity, including 109 spectral matches and 30 compounds with confirmed toxic activity.

Nontargeted screening (NTS) utilizing liquid chromatography electrospray ionization high-resolution mass spectrometry (LC/ESI/HRMS) is increasingly used to identify environmental contaminants. Major differences in the ionization efficiency of compounds in ESI/HRMS result in widely varying responses and complicate quantitative analysis. Despite an increasing number of methods for quantification without authentic standards in NTS, the approaches are evaluated on limited and diverse data sets with varying chemical coverage collected on different instruments, complicating an unbiased comparison. In this interlaboratory comparison, organized by the NORMAN Network, we evaluated the accuracy and performance variability of five quantification approaches across 41 NTS methods from 37 laboratories. Three approaches are based on surrogate standard quantification (parent-transformation product, structurally similar or close eluting) and two on predicted ionization efficiencies (RandFor-IE and MLR-IE). Shortly, HPLC grade water, tap water, and surface water spiked with 45 compounds at 2 concentration levels were analyzed together with 41 calibrants at 6 known concentrations by the laboratories using in-house NTS workflows. The accuracy of the approaches was evaluated by comparing the estimated and spiked concentrations across quantification approaches, instrumentation, and laboratories. The RandFor-IE approach performed best with a reported mean prediction error of 15× and over 83% of compounds quantified within 10× error. Despite different instrumentation and workflows, the performance was stable across laboratories and did not depend on the complexity of water matrices.