Molecule Property Prediction Based on Spatial Graph Embedding

Authors:

Abstract
Accurate prediction of molecular properties is important for new compounds design, which is a crucial step in drug discovery. In this paper, molecular graph data is utilized for properties prediction based on graph convolution neural net-work. In addition, a Convolution Spatial Graph Embedding Layer (C-SGEL) is introduced to retain the spatial connection information of molecules. And multiple C-SGELs are stack to construct a Convolution Spatial Graph Embedding Network (C-SGEN) for end-to-end representation learning. In order to enhance the robustness of the network, molecular finger-prints are also combined with C-SGEN to build a composite model for predicting molecular properties. Our comparative experiments have shown that our method is accurate, and achieves the best results on some open benchmark datasets.
Funding Information
  • National Natural Science Foundation of China (61872326, 61602430)

This publication has 31 references indexed in Scilit: