Wang Hongmei, Xu Ming, Guo Zhitong, You Guilin, Wang Guishen, Cao Chen, Hu Xiaowen
College of Computer Science and Engineering, Changchun University of Technology, Changchun, Jilin 130012, China.
College of Computer Science and Engineering, Changchun University of Technology, Changchun, Jilin 130012, China.
Comput Biol Chem. 2025 Aug;117:108410. doi: 10.1016/j.compbiolchem.2025.108410. Epub 2025 Mar 8.
Accurate prediction of Drug-Target Interactions (DTIs) is crucial for drug discovery and development. While current research focuses predominantly on modern medicine, we propose DTI-BGCGCN, a novel predictive model that integrates a bipartite drug-target attribute graph with a Cluster Graph Con- volutional Network (ClusterGCN) for both modern and traditional Chinese medicine. Our approach employs a bipartite attribute graph to efficiently en- capsulate drug-target relationships and common features, while ClusterGCN classifies different graph topological structures and expedites the training process. Extensive experiments on both modern drug and traditional Chinese medicine datasets demonstrate that DTI-BGCGCN outperforms existing methodologies. Comprehensive ablation studies underscore the efficacy of key components within the framework. This approach presents a promising avenue for accelerating drug discovery through improved DTI prediction accuracy, bridging the gap between modern and traditional medicine in com- putational drug research.
准确预测药物-靶点相互作用（DTIs）对于药物发现和开发至关重要。虽然目前的研究主要集中在现代医学上，但我们提出了DTI-BGCGCN，这是一种新颖的预测模型，它将二分药物-靶点属性图与聚类图卷积网络（ClusterGCN）相结合，用于现代医学和传统中药。我们的方法采用二分属性图来有效地封装药物-靶点关系和共同特征，而ClusterGCN对不同的图拓扑结构进行分类并加快训练过程。在现代药物和传统中药数据集上进行的大量实验表明，DTI-BGCGCN优于现有方法。全面的消融研究强调了框架内关键组件的有效性。这种方法为通过提高DTI预测准确性来加速药物发现提供了一条有前途的途径，弥合了计算药物研究中现代医学和传统医学之间的差距。
