Predicting scalar coupling constants by graph angle-attention neural network

Abstract

Scalar coupling constant (SCC), directly measured by nuclear magnetic resonance (NMR) spectroscopy, is a key parameter for molecular structure analysis, and widely used to predict unknown molecular structure. Restricted by the high cost of NMR experiments, it is impossible to measure the SCC of unknown molecules on a large scale. Using density functional theory (DFT) to theoretically calculate the SCC of molecules is incredibly challenging, due to the cost of substantial computational time and space. Graph neural networks (GNN) of artificial intelligence (AI) have great potential in constructing molecul ar-like topology models, which endows them the ability to rapidly predict SCC through data-driven machine learning methods, and avoiding time-consuming quantum chemical calculations. With a priori knowledge of angles, we propose a graph angle-attention neural network (GAANN) model to predict SCC by means of some easily accessible related information. GAANN, with a multilayer message-passing network and a self-attention mechanism, can accurately simulate the molecular-like topological structure and predict molecular properties. Our simulations show that the prediction accuracy by GAANN, with the log(MAE) = −2.52, is close to that by DFT calculations. Different from conventional AI methods, GAANN combining the AI method with quantum chemistry theory (Karplus equation) has a strong physicochemical interpretability about angles. From an AI perspective, we find that bond angle has the highest correlation with the SCC among all angle features (dihedral angle, bond angle, geometric angles) about multiple coupling types in the small molecule datasets.

MA’AT: A Web-Based Application to Determine Rotamer Population Distributions in Solution from Nuclear Magnetic Resonance Spin-Coupling Constants

“…Density functional theory (DFT) has been used to obtain relationships between individual J-couplings and one or more molecular properties. 5,7,10,11 Current DFT calculations, molecular dynamics (MD) simulations, and other computational methods 12,13 are able to treat molecules that exist as conformational populations in solution. Conversely, determining conformational populations (equilibria) in solution from experimental data is not straightforward.…”

“…Current DFT calculations, molecular dynamics (MD) simulations, and other computational methods , are able to treat molecules that exist as conformational populations in solution. Conversely, determining conformational populations (equilibria) in solution from experimental data is not straightforward.…”

Performance prediction of magnetorheological fluid‐based liquid gating membrane by kriging machine learning method

“…Machine learning methods, including regression method, neural networks, support vector machine (SVM), clustering algorithm (K-Means), can build the relationship between inputs and outputs based on limited data, which is promising in predicting of material science, Miao et al [19] accelerated the discovery of new catalysts by combining active machine learning with density functional theory (DFT) calculations, and the prediction accuracy of the new neural network constructed by Jia et al [20] is close to the calculation results of DFT. In the field of MRF material predicting, Bahiuddin et al [21,22] developed an extreme learning machine method to predict the properties of the MRF constitutive model.…”