图神经网络在药物-靶标预测领域的研究进展  

Research Progress of Graph Neural Network in the Field of Drug-target Prediction

作　　者：孙连伟 连雪全 杨佳林 李建伟 SUN Lianwei;LIAN Xuequan;YANG Jialin;LI Jianwei(School of Artificial Intelligence,Hebei University of Technology,Tianjin 300401,China)

机构地区：[1]河北工业大学人工智能与数据科学学院,天津300401

出　　处：《医学信息》2025年第8期167-171,共5页Journal of Medical Information

基　　金：国家自然科学基金项目(编号:62072154)。

摘　　要：药物-靶标预测是现代药物研发中至关重要的一个环节,它克服了传统药物研发中发现和验证药物-靶标对的高成本和周期长等问题。药物-靶标相互作用(DTI)预测和药物-靶标亲和力(DTA)预测都是药物-靶标预测的任务。预测药物-靶标相互作用,筛选出具有潜在相互作用的药物-靶标对,对药物重定位以及新药研发具有重要意义。预测药物-靶标亲和力对药物的设计和优化可起到辅助性作用。图神经网络(GNN)作为一种用于处理图结构数据的深度学习模型,它可有效地对同构网络或异构网络中的节点和边进行表示学习,现已被广泛应用于各种药物-靶标预测任务中。本文对图神经网络在药物-靶标预测领域的研究进展作一综述,以期为图神经网络在药物研发领域的进一步发展提供参考。Drug-target prediction is a crucial part of modern drug research and development.It overcomes the problems of high cost and long cycle in the discovery and verification of drug-target pairs in traditional drug research and development.Drug-target interaction(DTI) prediction and drugtarget affinity(DTA) prediction are both tasks of drug-target prediction.Predicting drug-target interactions and screening drug-target pairs with potential interactions are of great significance for drug repositioning and new drug development.Predicting drug-target affinity can play an auxiliary role in drug design and optimization.As a deep learning model for processing graph-structured data,graph neural network(GNN) can effectively represent nodes and edges in homogeneous or heterogeneous networks,and has been widely used in various drug-target prediction tasks.This paper reviews the research progress of graph neural network in the field of drug-target prediction,in order to provide reference for the further development of graph neural network in the field of drug research and development.

关 键 词：药物-靶标预测 药物-靶标相互作用 药物-靶标亲和力 图神经网络 异构网络 

分 类 号：Q811[生物学—生物工程]