A cocrystal prediction method of graph neural networks based on molecular spatial information and global attention
Yanlei Kang, Jiahui Chen, Xiu-Rong Hu, Yunliang Jiang, Li Zhong
Abstract
This paper proposes a PFGNet model based on graph neural network, which adds molecular 3D point cloud information to represent the global aspect of molecules through the global attention mechanism.
Topics & Concepts
Computer scienceCocrystalArtificial neural networkGraphPoint cloudTheoretical computer scienceGlobal networkDeep neural networksArtificial intelligenceData miningMoleculeChemistryComputer networkHydrogen bondOrganic chemistryMachine Learning in Materials ScienceComputational Drug Discovery MethodsCrystallography and molecular interactions