郑州大学学报（理学版）2026，Vol.58Issue(4)：70-77,8.DOI:10.13705/j.issn.1671-6841.2024131
基于序列属性和结构特征的噬菌体-宿主多关系相互预测
A Multi-relationship Interaction Prediction Model of Phage and Host Based on Sequence and Structural Features
摘要
Abstract
As potential therapies for treating bacterial diseases,exploration on phage-host interactions(PHIs)can deepen the understanding of phage bactericidal mechanisms.In recent years,machine learning showned excellent performance in PHI prediction.However,it still faced challenges of data sparsity and missing biological attributes.To address this problem,a phage-host interaction prediction model(PHI-SAS)that combined biological sequence properties with network topological features was proposed.Firstly,A dataset containing 9 871 PHI pairs was constructed.Secondly,A multi-relationship struct aware module(MRSAM)based on generative adversarial networks and a multi-feature fusion pre-diction method(MFFPM)to maximize the complementarity between different features were designed.Ex-perimental results demonstrated that when evaluated using 5-fold cross-validation,PHI-SAS could provide higher prediction accuracy and robustness in predicting phage-host interactions compared to existing mod-els.关键词
噬菌体-宿主相互作用/属性缺失/网络嵌入/生成对抗网络/特征融合Key words
phage-host interactions/attribute missing/network embedding/generative adversarial network/feature fusion分类
信息技术与安全科学引用本文复制引用
王文,许文俊,陈诚,夏迎春,王庆勇,辜丽川..基于序列属性和结构特征的噬菌体-宿主多关系相互预测[J].郑州大学学报（理学版）,2026,58(4):70-77,8.基金项目
国家自然科学基金项目(62301006) （62301006）
国家重点研发计划项目(2023YFD1802200) （2023YFD1802200）
