Boosting Graph Neural Networks with Molecular Mechanics: A Case Study of Sigma Profile Prediction

JCTC Early Career Board Selects

“…Zhao et al demonstrated that simple classifiers can learn the features of reaction conformers that lead to successful transition state searches, accelerating reaction mechanism characterization across over three hundred individual reactions . Abranches et al developed a graph convolutional network framework to predict sigma profiles from molecular structures, providing a cost-effective alternative to density functional theory calculations to screen the physicochemical properties of large molecular data sets . Hruska et al developed machine learning models to reduce the errors of quantum mechanical redox potential calculations in both implicit and explicit solvent models relative to experimental measurements .…”