基于全局图注意力元路径异构网络的药物-疾病关联预测  

Drug-Disease Association Prediction Based on Meta-Path Heterogeneous Network with Global Graph Attention

作　　者：郁湧[1,2] 杨雨洁 李虓晗 高悦 于倩 YU Yong;YANG Yujie;LI Xiaohan;GAO Yue;YU Qian(School of Software,Yunnan University,Kunming 650504,China;Key Laboratory in Software Engineering of Yunnan Province,Kunming 650504,China)

机构地区：[1]云南大学软件学院,昆明650504 [2]云南省软件工程重点实验室,昆明650504

出　　处：《电子科技大学学报》2024年第4期576-583,共8页Journal of University of Electronic Science and Technology of China

基　　金：国家自然科学基金(62366058);云南省科技厅面上项目(202001BB050063);云南省软件工程重点实验室开放项目(2020SE315)。

摘　　要：提出了一个基于全局图注意力元路径异构网络模型(MHNGA)来进行药物-疾病关联预测。首先,收集整理药物和疾病数据,将已知的药物-疾病关联、药物相似性、疾病相似性构建为一个异构网络;其次,引入多个基于元路径的子图,使用图注意力神经网络提取这些子图的邻居节点的特征,并且通过通道注意力和空间注意力机制来增强特征;最后,通过十折交叉验证的评估,MHNGA取得了93.5%的精确召回曲线下的面积和99.4%的准确率。In this paper,a heterogeneous network model based on global graph attention meta-path,named MHNGA,is proposed for drug-disease association prediction.Firstly,the data of drugs and diseases are collected,and the known drug-disease association,drug similarity and disease similarity are constructed as a heterogeneous network.Secondly,multiple meta-path-based subgraphs are introduced,and the graph attention neural network is used to extract the features of the neighbor nodes of these subgraphs,and the features are enhanced by channel attention and spatial attention mechanisms.Finally,through the evaluation of ten-fold cross-validation,MHNGA achieves 93.5%of the area under the accurate recall curve and 99.4%of the accuracy.

关 键 词：异构图 药物-疾病关联 预测 图注意力神经网络 元路径 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]