School of Informatics Science and Technology, Zhejiang Sci-Tech University, Hangzhou 310018, China.
Zhejiang Province Key Laboratory of Smart Management & Application of Modern Agricultural Resources, School of Information Engineering, Huzhou University, Huzhou 313000, China.
Biomolecules. 2022 Nov 10;12(11):1666. doi: 10.3390/biom12111666.
Drug repositioning, an important method of drug development, is utilized to discover investigational drugs beyond the originally approved indications, expand the application scope of drugs, and reduce the cost of drug development. With the emergence of increasingly drug-disease-related biological networks, the challenge still remains to effectively fuse biological entity data and accurately achieve drug-disease repositioning. This paper proposes a new drug repositioning method named EMPHCN based on enhanced message passing and hypergraph convolutional networks (HGCN). It firstly constructs the homogeneous multi-view information with multiple drug similarity features and then extracts the intra-domain embedding of drugs through the combination of HGCN and channel attention mechanism. Secondly, inter-domain information of known drug-disease associations is extracted by graph convolutional networks combining node and edge embedding (NEEGCN), and a heterogeneous network composed of drugs, proteins and diseases is built as an important auxiliary to enhance the inter-domain message passing of drugs and diseases. Besides, the intra-domain embedding of diseases is also extracted through HGCN. Ultimately, intra-domain and inter-domain embeddings of drugs and diseases are integrated as the final embedding for calculating the drug-disease correlation matrix. Through 10-fold cross-validation on some benchmark datasets, we find that the AUPR of EMPHCN reaches 0.593 (T1) and 0.526 (T2), respectively, and the AUC achieves 0.887 (T1) and 0.961 (T2) respectively, which shows that EMPHCN has an advantage over other state-of-the-art prediction methods. Concerning the new disease association prediction, the AUC of EMPHCN through the five-fold cross-validation reaches 0.806 (T1) and 0.845 (T2), which are 4.3% (T1) and 4.0% (T2) higher than the second best existing methods, respectively. In the case study, EMPHCN also achieves satisfactory results in real drug repositioning for breast carcinoma and Parkinson's disease.
药物重定位是一种重要的药物开发方法，用于发现超出最初批准适应症的研究药物，扩大药物的应用范围，并降低药物开发成本。随着越来越多的药物-疾病相关生物网络的出现，仍然需要有效地融合生物实体数据并准确实现药物-疾病重新定位。本文提出了一种新的药物重定位方法，名为 EMPHCN，基于增强消息传递和超图卷积网络（HGCN）。它首先构建具有多个药物相似性特征的同质性多视图信息，然后通过 HGCN 和通道注意力机制的组合提取药物的域内嵌入。其次，通过结合节点和边嵌入（NEEGCN）的图卷积网络提取已知药物-疾病关联的域间信息，并构建一个由药物、蛋白质和疾病组成的异质网络，作为增强药物和疾病域间消息传递的重要辅助。此外，还通过 HGCN 提取疾病的域内嵌入。最终，将药物和疾病的域内和域间嵌入整合为计算药物-疾病相关矩阵的最终嵌入。通过在一些基准数据集上进行 10 折交叉验证，我们发现 EMPHCN 的 AUPR 在 T1 和 T2 上分别达到 0.593 和 0.526，AUC 在 T1 和 T2 上分别达到 0.887 和 0.961，这表明 EMPHCN 优于其他最先进的预测方法。关于新的疾病关联预测，通过五折交叉验证的 EMPHCN 的 AUC 在 T1 和 T2 上分别达到 0.806 和 0.845，分别比第二好的现有方法高 4.3%（T1）和 4.0%（T2）。在案例研究中，EMPHCN 在乳腺癌和帕金森病的实际药物重定位中也取得了令人满意的结果。
