AGraphDTA: An Efficient Model for Drug-Target Affinity Prediction with Feature Fusion
Donglin Wang, Xiangyong Chen, Xin Bao, Kun Zhou · 2023
It is a crucial task to predict the Drug-target affinity (DTA) in drug discovery. Recently, the application of deep learning shows a significant improvement on DTA prediction. However, most previous methods have struggled to extract the complex information of proteins. Hence, this paper introduces a novel model named AGraphDTA for DTA prediction. Specifically, AGraphDTA employs Graph Neural Network (GNN) to extract graph features of drugs and proteins. Then it employs Convo-lutional Neural Network (CNN) to extract amino acid sequence features. Additionally, a fusing block is employed to generate the fusion feature that represent the complex information of proteins. Evaluation results on benchmark datasets show that AGraphDTA demonstrates a superior performance in DTA prediction.