Abstract:Developing quantitative structure-activity relationships (QSARs) is an important approach to predicting the reactivity of HO radicals toward newly emerged organic compounds. As compared with molecular descriptors-based and the group contribution method-based QSARs, a combined molecular fingerprint-machine learning (ML) method can more quickly and accurately develop such models for a growing number of contaminants. However, it is yet unknown whether this method makes predictions by choosing meaningful structural features rather than spurious ones, which is vital for trusting the models. In this study, we developed QSAR models for the logkHO values of 1089 organic compounds in the aqueous phase by two ML algorithms—deep neural networks (DNN) and eXtreme Gradient Boosting (XGBoost), and interpreted the built models by the SHapley Additive exPlanations (SHAP) method. The results showed that for the contribution of a given structural feature to logkHO for different compounds, DNN and XGBoost treated it as a fixed and variable value, respectively. We then developed an ensemble model combining the DNN with XGBoost, which achieved satisfactory predictive performance for all three datasets: Training dataset: R-square (R2) 0.89–0.91, root-mean-squared-error (RMSE) 0.21–0.23, and mean absolute error (MAE) 0.15–0.17; Validation dataset: R2 0.63–0.78, RMSE 0.29–0.32, and MAE 0.21–0.25; and Test dataset: R2 0.60–0.71, RMSE 0.30–0.35, and MAE 0.23–0.25. The SHAP method was further used to unveil that this ensemble model made predictions on logkHO based on a correct 'understanding' of the impact of electron-withdrawing and -donating groups and of the reactive sites in the compounds that can be attacked by HO. This study offered some much-needed mechanistic insights into a ML-assisted environmental task, which are important for evaluating the trustworthiness of the ML-based models, further improving the models for specific applications, and leveraging the implicit knowledge the models carry.

Predicting Rate Constants of Hydroxyl Radical Reactions with Alkanes Using Machine Learning

Prediction of Hydrogen Abstraction Rate Constants at the Allylic Site between Alkenes and OH with Multiple Machine Learning Models

Predicting Rate Constants of Alkane Cracking Reactions Using Machine Learning

Rate coefficients for the reactions of OH radicals with C3–C11 alkanes determined by the relative-rate technique

Theoretical Study of an Undisclosed Reaction Class: Direct H-Atom Abstraction from Allylic Radicals by Molecular Oxygen

Machine Learning Prediction of Hydrogen Atom Transfer Reactivity in Photoredox-Mediated C–H Functionalization

Predicting the Enthalpy of Hydrocarbon Radicals Adsorbed on Pt(111) Using Molecular Fingerprints and Machine Learning

Machine learning for predicting halogen radical reactivity toward aqueous organic chemicals

An ONIOM-Based High-Level Thermochemistry Study on Hydrogen Abstraction Reactions of Large Straight-Chain Alkanes by Hydrogen, Hydroxyl, and Hydroperoxyl Radicals

Theoretical Investigation on the H Atom Abstraction Reaction from C1-C4 Alkanes and Alkenes by ṄH2 Radicals

From electronic structure to model application for alkyl cyclohexane combustion chemistry: H-atom abstraction reactions by HȮ2 radical

Comprehensive Multipath Variational Kinetics Study on Hydrogen Abstraction Reactions from Three Typical Dimethylcyclohexane Isomers by Hydroxyl Radicals: from the Electronic Structure to Model Applications

Theoretical kinetic study of key reactions between ammonia and fuel molecules, part I: Hydrogen atom abstraction from alkanes by ṄH 2 radical

From electronic structure to model application of key reactions for gasoline/alcohol combustion: Hydrogen-atom abstractions by CH 3 Ȯ radical

Prediction of Free Radical Reactions Toward Organic Pollutants with Easily Accessible Molecular Descriptors.

From electronic structure to model application of key reactions for gasoline/alcohol combustion: Hydrogen-atom abstraction by CH 3 OȮ radicals

Predictive Combustion Kinetics of OH Radical Reactions with a C5 Unsaturated Alcohol: The Competitive H-Abstraction and OH-Addition Reactions of 2-Methyl-3-buten-2-ol

Multi-structural variational kinetics study on hydrogen abstraction reactions of cyclopentanol and cyclopentane by hydroperoxyl radical with anharmonicity, recrossing and tunneling effects

Predicting Regioselectivity in Radical C−H Functionalization of Heterocycles Through Machine Learning

Shedding light on “Black Box” machine learning models for predicting the reactivity of HO radicals toward organic compounds

Understanding kinetic interactions between NOx and C2-C5 alkanes and alkenes: The rate rules and influences of H-atom abstractions by NO2