Molecular docking method based on scoring device and dynamic graph network

• Main Authors: GUO TAO, DENG YUE
• Format: Patent
• Language: Chinese; English
• Published: 15.07.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTEDFOR SPECIFIC APPLICATION FIELDS; PHYSICS

The invention discloses a molecular docking method based on a scoring device and a dynamic graph network, and the method comprises the following steps: constructing a graph neural network of a predictor, namely, defining node features and edge features; updating the graph neural network of the constructed predictor; predicting the affinity based on a predictor; constructing and training a loss function of the predictor; constructing a generator, and defining a log-likelihood function; predicting the gradient of the log-likelihood function based on the generator; training the generator, and performing conformation optimization based on Langevin sampling; and completing molecular docking based on the predictor and the generator. According to the method, the graph neural network is utilized to realize docking of the small molecules and the proteins, the graph neural network is dynamically constructed to perform characterization learning of the proteins and the small molecules, and the graph neural network is uti