Research On Drug-target Affinity Prediction Method Based On Deep Learnin

Drug discovery is a complicated process,which has the risks of tall research cycles,high costs,and low success rates.It takes billions of dollars and more than ten years to develop a new drug from development to approval.Although the investment is high and the time is long,the success rate of the final marketing of small molecule drugs is only13%,and the risk of failure is high.The effective prediction of drug-target binding affinity(DTA)is one of the significant issues in drug discovery.However,limited by human and material resources,traditional biological experiments are difficult to achieve large-scale and high-throughput screening.Benefiting from the development of information technology,the method of computer-aided prediction of drug-target binding affinity has been realized and widely used.Computer-aided drug-target binding affinity prediction guides the discovery and modification of lead compounds,reducing blindness in drug development and speeding up the process.In addition,drug-target binding affinity prediction can identify potential protein targets for existing drugs,thus discovering their new indications.Therefore,it is very important to develop efficient and accurate prediction algorithms.The key to drug-target binding affinity prediction based on deep learning lies in accurate and complete information representation and feature extraction of drug and target.In view of the existing problems of drug and protein representation in the existing algorithm,this paper mainly does three works:(1)This paper proposes a drug-target affinity prediction method based on attention mechanism,which enhances the model’s learning ability to improve prediction performance.Unlike traditional methods,this method uses self-attention mechanism to extract effective drug features on the drug molecular graph,and designs target attention mechanism for protein features.The target attention mechanism can learn the importance score of protein information after the convolutional neural network captures the potential features of protein sequence.Finally,the drug features and protein features obtained through attention mechanism are input to the fully connected layer for affinity prediction.(2)This paper proposes a method for predicting drug-target affinity based on a fusion graph neural network and interactive attention.The method uses the structural information of molecules and proteins to construct a molecular graph for drugs and a contact graph for proteins.By combining graph convolutional neural networks and Transformers,the method uses a fusion graph neural network to learn global information and dependencies between nodes in the molecular and protein graphs.In addition,the method designs an interactive attention mechanism to model the interaction information between drugs and targets,integrating the interaction information between proteins and drugs.(3)This paper proposes a novel model based on Transformer and Graph-Sequence attention to predict the binding affinity between drugs and targets.For the representation of drugs,we use Bi-directional Gated Recurrent Units(Bi GRU)to extract the SMILES representation from SMILES sequences,and graph neural networks to extract the graph representation of the molecular graphs.Then we utilize an attention mechanism to fuse the two representations of the drug.For the target/protein,we utilized the Efficient Transformer to learn the representation of protein,which can capture the long-distance relationship in the sequence of amino acids.