%0 Journal Article
%T 基于图神经网络预测药物&#8211;靶标相互作用的方法综述<br>A Review of Methods for Predicting Drug-Target Interactions Based on Graph Neural Networks
%A 左乐
%A 张琪
%J Hans Journal of Biomedicine
%P 563-572
%@ 2161-8984
%D 2024
%I Hans Publishing
%R 10.12677/hjbm.2024.144060
%X 药物&#8211;靶标相互作用(DTI)在药物发现中至关重要&#65292;利用人工智能模型对DTI进行预测不仅减少了时间和成本&#65292;还能够提高预测的准确性和精度。然而&#65292;DTI数据异质性的特点以及预测模型的可解释性为研究人员带来挑战。在目前的研究中&#65292;图神经网络(GNN)在分析和处理多源数据挖掘其中的复杂关系及可解释性方面具有显著的优势&#65292;是DTI预测研究的强大工具。文章从三个方面对GNN在DTI预测中的方法进行综述。首先&#65292;介绍了DTI预测的研究现状。其次&#65292;分别探讨了GNN在药物(靶标)特征提取和DTI预测的最新研究进展&#65292;整理了常用的公开数据集。最后&#65292;总结了基于GNN预测DTI研究中存在的挑战并为未来相关研究提供展望。<br>Drug-target interaction (DTI) is crucial in drug discovery, and utilizing artificial intelligence models to predict DTI not only reduces time and costs but also enhances prediction accuracy and precision. However, the heterogeneity of DTI data and the interpretability of prediction models pose challenges for researchers. In current research, graph neural networks (GNNs) exhibit significant advantages in analyzing and processing complex relationships among multiple data sources and enhancing interpretability, making them powerful tools in DTI prediction studies. This article provides a comprehensive review of GNN methods in DTI prediction from three aspects. Firstly, it introduces the current status of DTI prediction research. Secondly, it discusses the latest research progress of GNNs in drug (target) feature extraction and DTI prediction, and compiles commonly used public datasets. Finally, it summarizes the challenges in DTI prediction research based on GNNs and provides prospects for future related studies.
%K 图神经网络&#65292
%K 药物与靶标相互作用&#65292
%K 异构网络<br>Graph Neural Networks
%K Drug-Target Interactions
%K Heterogeneous Networks
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=97846