Behavior-aware contextual representation learning for drug-target interaction prediction

Yimin Yin; Jianying Qu; Yingying Li; Jing Zhang; Bin Yan; Wanxia Deng; Jinghua Zhang; Yimin Yin; Jianying Qu; Yingying Li; Jing Zhang; Bin Yan; Wanxia Deng; Jinghua Zhang

doi:10.3934/era.2026339

Electronic Research Archive

2026, Volume 34, Issue 11: 7914-7936. doi: 10.3934/era.2026339

Previous Article Next Article

Research article

Behavior-aware contextual representation learning for drug-target interaction prediction

1.
School of Mathematics and Statistics, Hunan First Normal University, Changsha 410205, China
2.
Hunan Provincial University Key Laboratory for Big Data Analysis and Application, Hunan First Normal University, Changsha 410205, China
3.
College of Science, National University of Defense Technology, Changsha 410073, China
4.
Key Laboratory of Computing and Stochastic Mathematics, Hunan Normal University, Changsha 410081, China
5.
College of Meteorology and Oceanography, National University of Defense Technology, Changsha 410073, China
6.
College of Computer Science and Software Engineering, Hohai University, Nanjing 211100, China

Received: 26 May 2026 Revised: 02 September 2026 Accepted: 03 September 2026 Published: 21 September 2026

Accurate prediction of drug-target interactions (DTIs) is a fundamental task in computational drug discovery, as it can effectively reduce the cost and time required for experimental screening. However, although most existing methods consider both sides of an interaction, they still focus mainly on the intrinsic features of drugs and proteins and are therefore insufficient to capture higher-order behavioral context and complementary interaction patterns in DTIs. To address this issue, we propose a novel DTI prediction framework that integrates drug molecular structure, protein sequence information, and behavior-aware contextual representations within a unified architecture. Specifically, drug compounds are encoded by a graph neural network (GNN) to capture molecular topology, protein sequences are modeled to learn semantic features, and a protein-centric behavioral sequence is introduced to characterize pharmacological interaction patterns. In addition, an attention-based fusion mechanism is introduced to model dependencies within the behavioral sequence and selectively aggregate informative behavioral contextual signals. Extensive experiments on three benchmark datasets demonstrate that the proposed method consistently outperforms representative baseline models. These results indicate that the proposed framework provides an effective computational solution for DTI prediction.
- drug-target interaction prediction,
- graph neural networks,
- behavioral sequence,
- attention mechanism
Citation: Yimin Yin, Jianying Qu, Yingying Li, Jing Zhang, Bin Yan, Wanxia Deng, Jinghua Zhang. Behavior-aware contextual representation learning for drug-target interaction prediction[J]. Electronic Research Archive, 2026, 34(11): 7914-7936. doi: 10.3934/era.2026339

Related Papers:

Abstract

Accurate prediction of drug-target interactions (DTIs) is a fundamental task in computational drug discovery, as it can effectively reduce the cost and time required for experimental screening. However, although most existing methods consider both sides of an interaction, they still focus mainly on the intrinsic features of drugs and proteins and are therefore insufficient to capture higher-order behavioral context and complementary interaction patterns in DTIs. To address this issue, we propose a novel DTI prediction framework that integrates drug molecular structure, protein sequence information, and behavior-aware contextual representations within a unified architecture. Specifically, drug compounds are encoded by a graph neural network (GNN) to capture molecular topology, protein sequences are modeled to learn semantic features, and a protein-centric behavioral sequence is introduced to characterize pharmacological interaction patterns. In addition, an attention-based fusion mechanism is introduced to model dependencies within the behavioral sequence and selectively aggregate informative behavioral contextual signals. Extensive experiments on three benchmark datasets demonstrate that the proposed method consistently outperforms representative baseline models. These results indicate that the proposed framework provides an effective computational solution for DTI prediction.

References

[1]	D. Sun, W. Gao, H. Hu, S. Zhou, Why 90% of clinical drug development fails and how to improve it?, Acta Pharm. Sin. B, 12 (2022), 3049–3062. https://doi.org/10.1016/j.apsb.2022.02.002 doi: 10.1016/j.apsb.2022.02.002
[2]	A. Schuhmacher, M. Hinder, E. Brief, O. Gassmann, D. Hartl, Benchmarking R&D success rates of leading pharmaceutical companies: an empirical analysis of FDA approvals (2006–2022), Drug Discov. Today, 30 (2025), 104291. https://doi.org/10.1016/j.drudis.2025.104291 doi: 10.1016/j.drudis.2025.104291
[3]	N. Q. K. Le, T. X. Tran, P. A. Nguyen, T. T. Ho, V. N. Nguyen, Recent progress in machine learning approaches for predicting carcinogenicity in drug development, Expert Opin. Drug Metab. Toxicol., 20 (2024), 621–628. https://doi.org/10.1080/17425255.2024.2356162 doi: 10.1080/17425255.2024.2356162
[4]	M. Korn, C. Ehrt, F. Ruggiu, M. Gastreich, M. Rarey, Navigating large chemical spaces in early-phase drug discovery, Curr. Opin. Struct. Biol., 80 (2023), 102578. https://doi.org/10.1016/j.sbi.2023.102578 doi: 10.1016/j.sbi.2023.102578
[5]	Q. Yao, Z. Chen, Y. Cao, H. Hu, Enhancing drug-target interaction prediction with graph representation learning and knowledge-based regularization, Front. Bioinf., 5 (2025), 1649337. https://doi.org/10.3389/fbinf.2025.1649337 doi: 10.3389/fbinf.2025.1649337
[6]	A. Gaulton, L. J. Bellis, A. P. Bento, J. Chambers, M. Davies, A. Hersey, et al., ChEMBL: a large-scale bioactivity database for drug discovery, Nucleic Acids Res., 40 (2012), D1100–D1107. https://doi.org/10.1093/nar/gkr777 doi: 10.1093/nar/gkr777
[7]	UniProt Consortium, UniProt: a worldwide hub of protein knowledge, Nucleic Acids Res., 47 (2019), D506–D515. https://doi.org/10.1093/nar/gky1049 doi: 10.1093/nar/gky1049
[8]	UniProt Consortium, UniProt: a hub for protein information, Nucleic Acids Res., 43 (2015), D204–D212. https://doi.org/10.1093/nar/gku989 doi: 10.1093/nar/gku989
[9]	M. Talo, S. Bozdag, Top-DTI: integrating topological deep learning and large language models for drug–target interaction prediction, Bioinformatics, 41 (2025), i133–i141. https://doi.org/10.1093/bioinformatics/btaf183 doi: 10.1093/bioinformatics/btaf183
[10]	F. Gentile, V. Agrawal, M. Hsing, A. T. Ton, F. Ban, U. Norinder, et al., Deep docking: a deep learning platform for augmentation of structure based drug discovery, ACS Cent. Sci., 6 (2020), 939–949. https://doi.org/10.1021/acscentsci.0c00229 doi: 10.1021/acscentsci.0c00229
[11]	W. Shi, H. Yang, L. Xie, X. X. Yin, Y. Zhang, A review of machine learning-based methods for predicting drug–target interactions, Health Inf. Sci. Syst., 12 (2024), 30. https://doi.org/10.1007/s13755-024-00287-6 doi: 10.1007/s13755-024-00287-6
[12]	Y. Ding, J. Tang, F. Guo, Q. Zou, Identification of drug–target interactions via multiple kernel-based triple collaborative matrix factorization, Briefings Bioinf., 23 (2022), bbab582. https://doi.org/10.1093/bib/bbab582 doi: 10.1093/bib/bbab582
[13]	F. Chen, Z. Zhao, Z. Ren, K. Lu, Y. Yu, W. Wang, Prediction of drug target interaction based on under sampling strategy and random forest algorithm, PLoS One, 20 (2025), e0318420. https://doi.org/10.1371/journal.pone.0318420 doi: 10.1371/journal.pone.0318420
[14]	R. Rodríguez-Pérez, J. Bajorath, Evolution of support vector machine and regression modeling in chemoinformatics and drug discovery, J. Comput.-Aided Mol. Des., 36 (2022), 355–362. https://doi.org/10.1007/s10822-022-00442-9 doi: 10.1007/s10822-022-00442-9
[15]	D. Song, Y. Chen, Q. Min, Q. Sun, K. Ye, C. Zhou, et al., Similarity-based machine learning support vector machine predictor of drug-drug interactions with improved accuracies, J. Clin. Pharm. Ther., 44 (2019), 268–275. https://doi.org/10.1111/jcpt.12786 doi: 10.1111/jcpt.12786
[16]	Y. C. Wang, Z. X. Yang, Y. Wang, N. Y. Deng, Computationally probing drug-protein interactions via support vector machine, Lett. Drug Des. Discov., 7 (2010), 370–378. https://doi.org/10.2174/157018010791163433 doi: 10.2174/157018010791163433
[17]	Q. Ning, Y. Wang, Y. Zhao, J. Sun, L. Jiang, K. Wang, et al., DMHGNN: Double multi-view heterogeneous graph neural network framework for drug-target interaction prediction, Artif. Intell. Med., 159 (2025), 103023. https://doi.org/10.1016/j.artmed.2024.103023 doi: 10.1016/j.artmed.2024.103023
[18]	Y. Jing, D. Zhang, L. Li, H2GnnDTI: hierarchical heterogeneous graph neural networks for drug–target interaction prediction, Bioinformatics, 41 (2025), btaf117. https://doi.org/10.1093/bioinformatics/btaf117 doi: 10.1093/bioinformatics/btaf117
[19]	O. Zhang, H. Lin, X. Zhang, X. Wang, Z. Wu, Q. Ye, et al., Graph neural networks in modern AI-aided drug discovery, Chem. Rev., 125 (2025), 10001–10103. https://doi.org/10.1021/acs.chemrev.5c00461 doi: 10.1021/acs.chemrev.5c00461
[20]	A. G. Vrahatis, K. Lazaros, S. Kotsiantis, Graph attention networks: a comprehensive review of methods and applications, Future Internet, 16 (2024), 318. https://doi.org/10.3390/fi16090318 doi: 10.3390/fi16090318
[21]	M. K. Subramanian, J. Spencer, Deep neural networks for in silico prediction of drug-target interactions: Advancements and future directions, Front. Emerg. Artif. Intell. Mach. Learn., 2 (2025), 1–12.
[22]	Z. Li, F. Liu, W. Yang, S. Peng, J. Zhou, A survey of convolutional neural networks: analysis, applications, and prospects, IEEE Trans. Neural Networks Learn. Syst., 33 (2021), 6999–7019. https://doi.org/10.1109/TNNLS.2021.3084827 doi: 10.1109/TNNLS.2021.3084827
[23]	A. Krizhevsky, I. Sutskever, G. E. Hinton, Imagenet classification with deep convolutional neural networks, Adv. Neural Inf. Process. Syst., 25 (2012).
[24]	K. He, X. Zhang, S. Ren, J. Sun, Deep residual learning for image recognition, in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), (2016), 770–778. https://doi.org/10.1109/CVPR.2016.90
[25]	A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, et al., Attention is all you need, Adv. Neural Inf. Process. Syst., 30 (2017).
[26]	D. Guo, D. Yang, H. Zhang, J. Song, P. Wang, Q. Zhu, et al., DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning, Nature, 645 (2025), 633–638. https://doi.org/10.1038/s41586-025-09422-z doi: 10.1038/s41586-025-09422-z
[27]	H. Öztürk, A. Özgür, E. Ozkirimli, DeepDTA: deep drug–target binding affinity prediction, Bioinformatics, 34 (2018), 821–829. https://doi.org/10.1093/bioinformatics/bty593 doi: 10.1093/bioinformatics/bty593
[28]	I. Lee, J. Keum, H. Nam, DeepConv-DTI: Prediction of drug-target interactions via deep learning with convolution on protein sequences, PLoS Comput. Biol., 15 (2019), e1007129. https://doi.org/10.1371/journal.pcbi.1007129 doi: 10.1371/journal.pcbi.1007129
[29]	K. Wang, R. Zhou, Y. Li, M. Li, DeepDTAF: a deep learning method to predict protein–ligand binding affinity, Briefings Bioinf., 22 (2021), bbab072. https://doi.org/10.1093/bib/bbab072 doi: 10.1093/bib/bbab072
[30]	K. Huang, C. Xiao, L. M. Glass, J. Sun, MolTrans: molecular interaction transformer for drug–target interaction prediction, Bioinformatics, 37 (2021), 830–836. https://doi.org/10.1093/bioinformatics/btaa880 doi: 10.1093/bioinformatics/btaa880
[31]	Q. Zhao, H. Zhao, K. Zheng, J. Wang, HyperAttentionDTI: improving drug–protein interaction prediction by sequence-based deep learning with attention mechanism, Bioinformatics, 38 (2022), 655–662. https://doi.org/10.1093/bioinformatics/btab715 doi: 10.1093/bioinformatics/btab715
[32]	P. Bai, F. Miljković, B. John, H. Lu, Interpretable bilinear attention network with domain adaptation improves drug–target prediction, Nat. Mach. Intell., 5 (2023), 126–136. https://doi.org/10.1038/s42256-022-00605-1 doi: 10.1038/s42256-022-00605-1
[33]	N. Q. K. Le, Predicting emerging drug interactions using GNNs, Nat. Comput. Sci., 3 (2023), 1007–1008. https://doi.org/10.1038/s43588-023-00555-7 doi: 10.1038/s43588-023-00555-7
[34]	T. M. Nguyen, T. Nguyen, T. M. Le, T. Tran, Gefa: early fusion approach in drug-target affinity prediction, IEEE/ACM Trans. Comput. Biol. Bioinf., 19 (2021), 718–728. https://doi.org/10.1109/TCBB.2021.3094217 doi: 10.1109/TCBB.2021.3094217
[35]	L. Peng, X. Liu, L. Yang, L. Liu, Z. Bai, M. Chen, et al., BINDTI: a bi-directional intention network for drug-target interaction identification based on attention mechanisms, IEEE J. Biomed. Health Inf., 29 (2024), 1602–1612. https://doi.org/10.1109/JBHI.2024.3375025 doi: 10.1109/JBHI.2024.3375025
[36]	A. Xiong, Z. Luo, Y. Xia, Q. Zou, L. Wei, Z. Zhang, et al., An interpretable geometric graph neural network for enhancing the generalizability of drug–target interaction prediction, BMC Biol., 23 (2025), 350. https://doi.org/10.1186/s12915-025-02456-9 doi: 10.1186/s12915-025-02456-9
[37]	J. Wei, Y. Zhu, L. Zhuo, Y. Liu, X. Fu, F. Li, Efficient deep model rnsemble framework for drug-target interaction prediction, J. Phys. Chem. Lett., 15 (2024), 7681–7693. https://doi.org/10.1021/acs.jpclett.4c01509 doi: 10.1021/acs.jpclett.4c01509
[38]	Z. Zhou, Q. Liao, J. Wei, L. Zhuo, X. Wu, X. Fu, et al., Revisiting drug-protein interaction prediction: a novel global-local perspective, Bioinformatics, 40 (2024), btae271. https://doi.org/10.1093/bioinformatics/btae271 doi: 10.1093/bioinformatics/btae271
[39]	R. Zhang, M. Yang, Q. Zhao, J. Wang, BiBLDR: Bidirectional behavior learning for drug repositioning, IEEE J. Biomed. Health Inf., (2025). https://doi.org/10.1109/JBHI.2025.3628673
[40]	M. K. Gilson, T. Liu, M. Baitaluk, G. Nicola, L. Hwang, J. Chong, BindingDB in 2015: a public database for medicinal chemistry, computational chemistry and systems pharmacology, Nucleic Acids Res., 44 (2016), D1045–D1053. https://doi.org/10.1093/nar/gkv1072 doi: 10.1093/nar/gkv1072
[41]	H. Liu, J. Sun, J. Guan, J. Zheng, S. Zhou, Improving compound–protein interaction prediction by building up highly credible negative samples, Bioinformatics, 31 (2015), i221–i229. https://doi.org/10.1093/bioinformatics/btv256 doi: 10.1093/bioinformatics/btv256
[42]	C. Cortes, V. Vapnik, Support-vector networks, Mach. Learn., 20 (1995), 273–297. https://doi.org/10.1007/BF00994018 doi: 10.1007/BF00994018
[43]	L. Breiman, Random forests, Mach. Learn., 45 (2001), 5–32. https://doi.org/10.1023/A: 1010933404324
[44]	T. Nguyen, H. Le, T. P. Quinn, T. Nguyen, T. D. Le, S. Venkatesh, GraphDTA: predicting drug–target binding affinity with graph neural networks, Bioinformatics, 37 (2021), 1140–1147. https://doi.org/10.1093/bioinformatics/btaa921 doi: 10.1093/bioinformatics/btaa921
[45]	L. Chen, X. Tan, D. Wang, F. Zhong, X. Liu, T. Yang, et al., TransformerCPI: improving compound–protein interaction prediction by sequence-based deep learning with self-attention mechanism and label reversal experiments, Bioinformatics, 36 (2020), 4406–4414. https://doi.org/10.1093/bioinformatics/btaa524 doi: 10.1093/bioinformatics/btaa524
[46]	X. Zeng, W. Chen, B. Lei, CAT-DTI: cross-attention and Transformer network with domain adaptation for drug-target interaction prediction, BMC Bioinf., 25 (2024), 141. https://doi.org/10.1186/s12859-024-05753-2 doi: 10.1186/s12859-024-05753-2
[47]	Y. Zhang, J. Li, S. Zhao, H. Tan, H. Sun, Y. Xiong, et al., Prediction of drug-target interactions based on hypergraph neural networks with multimodal feature fusion, IEEE J. Biomed. Health Inf., (2025). https://doi.org/10.1109/JBHI.2025.3625158
[48]	M. Gao, D. Zhang, Y. Chen, Y. Zhang, Z. Wang, X. Wang, et al., Graphormerdti: a graph transformer-based approach for drug-target interaction prediction, Comput. Biol. Med., 173 (2024), 108339. https://doi.org/10.1016/j.compbiomed.2024.108339 doi: 10.1016/j.compbiomed.2024.108339
[49]	L. Lebreton, O. Curet, S. Gueddari, F. Mazouz, S. Bernard, C. Burstein, et al., Selective and Potent Monoamine Oxidase Type B Inhibitors: 2-Substituted 5-Aryltetrazoles Derivatives, J. Med. Chem., 38 (1995), 4786–4792. https://doi.org/10.1021/jm00024a006 doi: 10.1021/jm00024a006
[50]	I. Abdullah, C. F. Chee, Y. K. Lee, S. S. R. Thunuguntla, K. S. Reddy, K. Nellore, et al., Benzimidazole derivatives as potential dual inhibitors for PARP-1 and DHODH, Bioorg. Med. Chem., 23 (2015), 4669–4680. https://doi.org/10.1016/j.bmc.2015.05.051 doi: 10.1016/j.bmc.2015.05.051
[51]	T. Y. Tsai, M. S. Coumar, T. Hsu, H. P. Hsieh, C. H. Chien, C. T. Chen, et al., Substituted pyrrolidine-2, 4-dicarboxylic acid amides as potent dipeptidyl peptidase IV inhibitors, Bioorg. Med. Chem. Lett., 16 (2006), 3268–3272. https://doi.org/10.1016/j.bmcl.2006.03.037 doi: 10.1016/j.bmcl.2006.03.037
[52]	J. Fiorito, F. Saeed, H. Zhang, A. Staniszewski, Y. Feng, Y. I. Francis, et al., Synthesis of quinoline derivatives: discovery of a potent and selective phosphodiesterase 5 inhibitor for the treatment of Alzheimer's disease, Eur. J. Med. Chem., 60 (2013), 285–294. https://doi.org/10.1016/j.ejmech.2012.12.009 doi: 10.1016/j.ejmech.2012.12.009
[53]	National Center for Biotechnology Information, PubChem BioAssay: AID 1798284, 2026. Available from: https://pubchem.ncbi.nlm.nih.gov/bioassay/1798284.
[54]	E. Lo Piparo, H. Scheib, N. Frei, G. Williamson, M. Grigorov, C. J. Chou, Flavonoids for controlling starch digestion: structural requirements for inhibiting human $\alpha$-amylase, J. Med. Chem., 51 (2008), 3555–3561. https://doi.org/10.1021/jm800115x doi: 10.1021/jm800115x
[55]	M. I. Davis, J. P. Hunt, S. Herrgard, P. Ciceri, L. M. Wodicka, G. Pallares, et al., Comprehensive analysis of kinase inhibitor selectivity, Nat. Biotechnol., 29 (2011), 1046–1051. https://doi.org/10.1038/nbt.1990 doi: 10.1038/nbt.1990
[56]	W. C. Kang, J. McAuley, Self-attentive sequential recommendation, in 2018 IEEE International Conference on Data Mining (ICDM), (2018), 197–206. https://doi.org/10.1109/ICDM.2018.00035
[57]	F. Sun, J. Liu, J. Wu, C. Pei, X. Lin, W. Ou, et al., BERT4Rec: Sequential recommendation with bidirectional encoder representations from transformer, in Proceedings of the 28th ACM International Conference on Information and Knowledge Management, (2019), 1441–1450. https://doi.org/10.1145/3357384.3357895
[58]	S. Yang, W. Ma, P. Sun, Q. Ai, Y. Liu, M. Cai, et al., Sequential recommendation with latent relations based on large language model, in Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieva, (2024), 335–344. https://doi.org/10.1145/3626772.3657762
[59]	W. Zhao, Y. Yu, G. Liu, Y. Liang, D. Xu, X. Feng, et al., MSI-DTI: predicting drug-target interaction based on multi-source information and multi-head self-attention, Briefings Bioinf., 25 (2024), bbae238. https://doi.org/10.1093/bib/bbae238 doi: 10.1093/bib/bbae238
[60]	Y. Meng, Y. Wang, J. Xu, C. Lu, X. Tang, T. Peng, et al., Drug repositioning based on weighted local information augmented graph neural network, Briefings Bioinf., 25 (2024), bbad431. https://doi.org/10.1093/bib/bbad431 doi: 10.1093/bib/bbad431
[61]	S. Kim, J. Chen, T. Cheng, A. Gindulyte, J. He, S. He, et al., PubChem 2023 update, Nucleic Acids Res., 51 (2023), D1373–D1380. https://doi.org/10.1093/nar/gkac956 doi: 10.1093/nar/gkac956
[62]	X. Chen, N. N. Guan, Y. Z. Sun, J. Q. Li, J. Qu, MicroRNA-small molecule association identification: from experimental results to computational models, Briefings Bioinf., 21 (2018), 47–61. https://doi.org/10.1093/bib/bby098 doi: 10.1093/bib/bby098
[63]	Z. Zhou, L. Zhuo, X. Fu, J. Lv, Q. Zou, R. Qi, Joint masking and self-supervised strategies for inferring small molecule-miRNA associations, Mol. Ther. Nucleic Acids, 35 (2024), 102103. https://doi.org/10.1016/j.omtn.2023.102103 doi: 10.1016/j.omtn.2023.102103
[64]	Z. Zhou, Z. Du, X. Jiang, L. Zhuo, Y. Xu, X. Fu, et al., GAM-MDR: probing miRNA-drug resistance using a graph autoencoder based on random path masking, Briefings Funct. Genomics, 23 (2024), 475–483. https://doi.org/10.1093/bfgp/elae005 doi: 10.1093/bfgp/elae005

Reader Comments

Your name:*

Email:*
© 2026 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)