Molecular distance matrix prediction based on graph convolutional networks

TL;DR

This paper introduces DMGCN, a graph convolutional network model that predicts molecular distance matrices efficiently, outperforming traditional methods and improving molecular property prediction accuracy.

Contribution

The paper presents a novel graph convolutional network approach for molecular distance matrix prediction, demonstrating improved accuracy over existing models and methods.

Findings

DMGCN achieves lower MAE than DeeperGCN-DAGNN and RDKit.

Predicted distances enhance molecular property prediction.

Model offers a faster alternative to DFT calculations.

Abstract

Molecular structure has important applications in many fields. For example, some studies show that molecular spatial information can be used to achieve better prediction results when predicting molecular properties. However, traditional molecular geometry calculations, such as density functional theory (DFT), are time-consuming. In view of this, we propose a model based on graph convolutional networks to predict the pairwise distance between atoms, also called distance matrix prediction of the molecule(DMGCN). In order to indicate the effect of DMGCN model, the model is compared with the model DeeperGCN-DAGNN and the method of calculating molecular conformation in RDKit. Results show that the MAE of DMGCN is smaller than DeeperGCN-DAGNN and RDKit. In addition, the distances predicted by the DMGCN model and the distances calculated by the QM9 dataset are used to predict the molecular properties, thus showing the effectiveness of the distance predicted by the DMGCN model.