Drug-target interaction prediction via hierarchical gated attention and information bottleneck

Abstract

Accurate drug-target interaction (DTI) prediction is essential for drug repositioning and accelerating drug discovery. Deep learning methods have made remarkable progress over traditional biological experiments, yet existing models often fail to capture local node topologies and multi-view semantic dependencies simultaneously. Moreover, most methods rely on basic loss functions that cannot filter out redundant noise, hindering the learning of compact and discriminative node representations. In this work, we propose a DTI prediction framework that integrates a hierarchical gated multi-head attention (HGMA) mechanism with an information bottleneck (IB) strategy. HGMA adopts a two-layer architecture: the first layer performs weighted aggregation over semantic meta-paths, and the second layer fuses attention heads via an adaptive gating mechanism, enhancing drug and target representations. The IB module compresses inputs by removing task-irrelevant redundancy while preserving predictive information, improving discriminability and generalization. Extensive experiments show that our model consistently outperforms state-of-the-art methods in both accuracy and robustness.

Keywords

Complex biological systems / drug-target interaction / hierarchical gated mechanism / graph attention / information bottleneck

References

[1] Zhou G,Wang Y,Wu H.Drug repositioning based on a multiplex network by integrating disease, gene, and drug information.Curr Bioinf2023;18:266-75

[2] Palhamkhani F,Dehnad A,Razzaghi P.DeepCompoundNet: enhancing compound-protein interaction prediction with multimodal convolutional neural networks.J Biomol Struct Dyn2023;43:1414-23

[3] Hu L,Ren Z.SSELM-neg: spherical search-based extreme learning machine for drug-target interaction prediction.BMC Bioin Bioinform2023;24:38 PMCID:PMC9896467

[4] Shi W,Xie L,Zhang Y.A review of machine learning-based methods for predicting drug-target interactions.Health Inf Sci Syst2024;12:30 PMCID:PMC11014838

[5] Zitnik M,Wang B,Goldenberg A.Machine learning for integrating data in biology and medicine: Principles, practice, and opportunities.Inf Fusion2019;50:71-91 PMCID:PMC6242341

[6] Zhang R,Wang X,Cui W.MHTAN-DTI: metapath-based hierarchical transformer and attention network for drug-target interaction prediction.Brief Bioinform2023;24:bbad079

[7] Yang Y,Pan A.Identifying depression in Parkinson's disease by using combined diffusion tensor imaging and support vector machine.Front Neurol2022;13:878691 PMCID:PMC9251067

[8] Yang J,Zhang Z.NegStacking: drug-target interaction prediction based on ensemble learning and logistic regression.IEEE/ACM Trans Comput Biol Bioinform2021;18:2624-34

[9] De A.DTI based Alzheimer’s disease classification with rank modulated fusion of CNNs and random forest.Expert Syst Appl2021;169:114338

[10] Liu B,Vens C.Drug-target interaction prediction via an ensemble of weighted nearest neighbors with interaction recovery.Appl Intell2021;52:3705-27

[11] Yamanishi Y,Gutteridge A,Kanehisa M.Prediction of drug-target interaction networks from the integration of chemical and genomic spaces.Bioinformatics2008;24:i232-40 PMCID:PMC2718640

[12] Jacob L.Protein-ligand interaction prediction: an improved chemogenomics approach.Bioinformatics2008;24:2149-56 PMCID:PMC2553441

[13] Tishby N,Bialek W.The information bottleneck method.arXiv2000;arXiv:physics/0004057

[14] Luo Y,Zhou J.A network integration approach for drug-target interaction prediction and computational drug repositioning from heterogeneous information.Nat Commun2017;8:573 PMCID:PMC5603535

[15] Scarselli F,Tsoi AC,Monfardini G.The graph neural network model.IEEE Trans Neural Netw2009;20:61-80

[16] Wan F,Xiao A,Zeng J.NeoDTI: neural integration of neighbor information from a heterogeneous network for discovering new drug-target interactions.Bioinformatics2019;35:104-11

[17] Zhao T,Valsdottir LR,Peng J.Identifying drug-target interactions based on graph convolutional network and deep neural network.Brief Bioinform2021;22:2141-50

[18] Peng J,Guan J.An end-to-end heterogeneous graph representation learning-based framework for drug-target interaction prediction.Brief Bioinform2021;22:bbaa430

[19] Li J,Lv H,Wang Z.IMCHGAN: inductive matrix completion with heterogeneous graph attention networks for drug-target interactions prediction.IEEE/ACM Trans Comput Biol Bioinform2022;19:655-65

[20] Su Y,Wang F.AMGDTI: drug-target interaction prediction based on adaptive meta-graph learning in heterogeneous network.Brief Bioinform2024;25:bbad474 PMCID:PMC10749791

[21] Wang M,Guo L,Pan Y.DHGT-DTI: advancing drug-target interaction prediction through a dual-view heterogeneous network with GraphSAGE and graph transformer.J Pharm Anal2025;15:101336 PMCID:PMC12616060

[22] Zhang Z,Qi Y.Leveraging 3D molecular spatial visual information and multi-perspective representations for drug discovery.Adv Sci2025;13:e12453 PMCID:PMC12786381

[23] Qiao G,Li Y.Causal enhanced drug-target interaction prediction based on graph generation and multi-source information fusion.Bioinformatics2024;40:btae570 PMCID:PMC11639159

[24] Li Y,Gao X.Supervised graph co-contrastive learning for drug-target interaction prediction.Bioinformatics2022;38:2847-54

[25] Yao K,Li W.Semi-supervised heterogeneous graph contrastive learning for drug-target interaction prediction.Comput Biol Med2023;163:107199

[26] Wu M,Ning Q,Guo S.Dss-Dti: drug-target interaction prediction method based on dual spatiotemporal scales.SSRN2025;

[27] Song W,Han C,Zou Q.Drug-target interaction predictions with multi-view similarity network fusion strategy and deep interactive attention mechanism.Bioinformatics2024;40:btae346 PMCID:PMC11164831

[28] Velickovic P,Casanova A.Graph attention networks.arXiv2017;abs/1710.10903

[29] Wang X,Gupta A.Non-local neural networks. In 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR); 2018 Jun 18-23; Salt Lake City, UT, USA. IEEE; 2018. pp. 7794-803.

[30] Dosovitskiy A,Kolesnikov A.An image is worth 16x16 words: transformers for image recognition at scale.ArXiv2020;abs/2010.11929.

[31] Jumper J,Pritzel A.Highly accurate protein structure prediction with AlphaFold.Nature2021;596:583-9 PMCID:PMC8371605

[32] Xiong Z,Liu X.Pushing the boundaries of molecular representation for drug discovery with the graph attention mechanism.J Med Chem2019;63:8749-60

[33] Balvir SU,Borkar PS.Node2Vec and machine learning: a powerful duo for link prediction in social network.J Electr Syst2024;20:639-49

[34] Mikolov T,Chen K.Distributed representations of words and phrases and their compositionality.Adv Neural Inform Process Syst2013;arXiv:1310.4546.

[35] Singh B,Vijayvargiya A.Analyzing the impact of activation functions on the performance of the data-driven gait model.Results Eng2023;18:101029

[36] Balestriero R.Batch normalization explained.arXiv2022;arXiv:2209.14778

[37] Dubey SR,Chaudhuri BB.Activation functions in deep learning: a comprehensive survey and benchmark.Neurocomputing2022;503:92-108

[38] Shen K,Tan X.A study on relu and softmax in transformer.arXiv2023;arXiv:2302.06461

[39] Du Y,Peng Z.Gated attention fusion network for multimodal sentiment classification.Knowl Based Syst2022;240:108107

[40] Sun Q,Peng H.Graph structure learning with variational information bottleneck.AAAI2022;36:4165-74

[41] Wu T,Wang J.Rethinking kullback-leibler divergence in knowledge distillation for large language models.arXiv2024;arXiv:2404.02657

[42] Huang W,Yan G,Luo H.EEG-based motor imagery classification using convolutional neural networks with local reparameterization trick.Expert Syst Appl2022;187:115968

[43] Tian Y,Yao X,Yang Y.Predicting molecular properties based on the interpretable graph neural network with multistep focus mechanism.Brief Bioinform2023;24:bbac534

[44] Zheng Y,Zhang X,Li J.Predicting drug targets from heterogeneous spaces using anchor graph hashing and ensemble learning. In 2018 International Joint Conference on Neural Networks (IJCNN); 2018 Jul 8-13; Rio de Janeiro. IEEE; 2018. pp. 1-7.

[45] Fan L.Optimization of Nadam algorithm for image denoising based on convolutional neural network. In 2020 7th International Conference on Information Science and Control Engineering (ICISCE); 2020 Dec 18-20; Changsha, China. IEEE; 2020. pp. 957-61.

[46] Carrington AM,Fieguth PW.Deep ROC analysis and AUC as balanced average accuracy, for improved classifier selection, audit and explanation.IEEE Trans Pattern Anal Mach Intell2023;45:329-41

[47] Lai B.Accurate protein function prediction via graph attention networks with predicted structure information.Brief Bioinform2022;23:bbab502 PMCID:PMC8898000

[48] Zheng Y,Wei Z.A survey of dynamic graph neural networks.Front Comput Sci2024;19:196323