Neural Message Passing for NMR Chemical Shift Prediction
Journal
J. Chem. Inf. Model. (Journal of Chemical Information and Modeling)
Date
2020.04.06
Abstract
Fast and accurate prediction of NMR spectra enables automatic structure validation and elucidation of molecules on a large-scale. In this paper, we propose an improved learning method to build a prediction model that learns from NMR database to predict the chemical shifts of NMR-active atoms for a new molecule. We adopt a message passing neural network that operates on the graph representation of a molecule. The graph representation is improved in terms of compactness and informativeness by treating hydrogen atoms implicitly and incorporating various node and edge features. The
proposed method demonstrates higher prediction performance for the 1H NMR and 13C NMR spectra of small molecules. We apply this method to molecule search, identifying the correct molecular structure of a new NMR spectrum by searching from a given set of candidate molecules.