Quantum chemistry-augmented neural networks for reactivity prediction: Performance, generalizability and interpretability

Abstract: There is a perceived dichotomy between structure-based and descriptor-based molecular representations used for predictive chemistry tasks. Here, we study the performance, generalizability, and interpretability of the recently proposed quantum mechanics-augmented graph neural network (ml-QM-GNN) architecture as applied to the prediction of regioselectivity (classification) and of activation energies (regression). In our hybrid QM-augmented model architecture, structure-based representations are first used to predict a set of atom- and bond-level reactivity descriptors derived from density functional theory (DFT) calculations. These estimated reactivity descriptors are combined with the original structure-based representation to make the final reactivity prediction. We demonstrate that our model architecture leads to significant improvements over structure-based GNNs in not only overall accuracy, but also in generalization to unseen compounds. Even when provided training sets of only a couple hundred labeled data points, the ml-QM-GNN outperforms other state-of-the-art model architectures that have been applied to these tasks. Further, because the predictions of our model are grounded in (but not restricted to) QM descriptors, we are able to relate predictions to the conceptual frameworks commonly used to gain qualitative insights into reactivity phenomena. This effort results in a productive synergy between theory and data science, wherein our QM-augmented models provide a data-driven confirmation of previous qualitative analyses, and these analyses in their turn facilitate insights into the decision-making process occurring within ml-QM-GNNs.