Abstract
Accurate and robust prediction of drug-target affinity (DTA) plays a critical role in drug discovery. While deep learning has advanced DTA prediction, existing methods struggle with limited training data and poor generalization. In this study, we propose AdaMBind, a novel DTA prediction model based on meta-learning framework with an adaptive task module designed for low-data scenarios. It employs a dynamic “easy-to-hard” task scheduling mechanism to enhance training efficiency and robustness. Experimental results on three benchmark datasets demonstrate that AdaMBind outperforms 8 baseline models in predicting affinity for unseen targets, particularly under few-shot conditions. Under stringent data constraints, the model successfully identifies high-affinity compounds for ESR and TP53, achieving outstanding virtual screening performance. Furthermore, when applied to inhibitor discovery against FLT3 for acute myeloid leukemia, AdaMBind successfully identified candidate compounds with potent inhibitory activity, as verified by preliminary experimental assays. In summary, AdaMBind provides a robust framework for few-shot DTA prediction.
Similar content being viewed by others
Data availability
The source data of three datasets used to train and evaluate the model is provided in https://github.com/Moohyun-w/AdaMBind/tree/main/data. The source data of the LIT-PCBA dataset is provided in https://drugdesign.unistra.fr/LIT-PCBA/. The support set data used to construct the FLT3 inhibitor prediction task are available at https://doi.org/10.5281/zenodo.1363539366. Source data are provided with this paper through https://doi.org/10.6084/m9.figshare.30963823. Source data are provided in this paper.
Code availability
The source code and data of this study are available at https://github.com/Moohyun-w/AdaMBind. The specific version of the code associated with this publication is archived in Zenodo and is accessible via 10.5281/zenodo.1859508467. Data are analyzed using numpy v2.2.6 (https://numpy.org/), pandas v2.3.2 (https://pandas.pydata.org/), Seaborn V0.13.2 (https://seaborn.pydata.org/). Structures are visualized by Pymol v3.1.6 (https://www.pymol.org/) and LigPlot68 v2.1 (https://www.ebi.ac.uk/). Molecular docking simulations are performed using AutoDock4 v4.2.6 (https://autodock.scripps.edu/).
References
Ijzerman, A. P. & Guo, D. Drug–target association kinetics in drug discovery. Trends Biochem. Sci. 44, 861–871 (2019).
Vamathevan, J. et al. Applications of machine learning in drug discovery and development. Nat. Rev. Drug Discov. 18, 463–477 (2019).
Ejalonibu, M. A. et al. Drug discovery for Mycobacterium tuberculosis using structure-based computer-aided drug design approach. Int. J. Mol. Sci. 22, 13259 (2021).
Thafar, M. et al. Comparison study of computational prediction tools for drug-target binding affinities. Front. Chem. 7, 782 (2019).
Mak, K. K. & Pichika, M. R. Artificial intelligence in drug development: present status and future prospects. Drug Discov. Today 24, 773–780 (2019).
Pinzi, L. & Rastelli, G. Molecular docking: shifting paradigms in drug discovery. Int. J. Mol. Sci. 20, 4331 (2019).
Chen, X. et al. Drug–target interaction prediction: databases, web servers and computational models. Brief. Bioinform. 17, 696–712 (2016).
Hua, Y. et al. MFR-DTA: a multi-functional and robust model for predicting drug–target binding affinity and region. Bioinformatics 39, btad056 (2023).
Li, T., Zhao, X. M. & Li, L. Co-VAE: Drug-target binding affinity prediction by co-regularized variational autoencoders. IEEE Trans. Pattern Anal. Mach. Intell. 44, 8861–8873 (2021).
Li, Z. et al. TEFDTA: a transformer encoder and fingerprint representation combined prediction method for bonded and non-bonded drug–target affinities. Bioinformatics 40, btad778 (2024).
Koh, H. Y. et al. Physicochemical graph neural network for learning protein–ligand interaction fingerprints from sequence data. Nat. Mach. Intell. 6, 673–687 (2024).
Fang, K. et al. ColdDTA: utilizing data augmentation and attention-based feature fusion for drug-target binding affinity prediction. Comput. Biol. Med. 164, 107372 (2023).
Zhou, C. et al. TransVAE-DTA: Transformer and variational autoencoder network for drug-target binding affinity prediction. Comput. Methods Programs Biomed. 244, 108003 (2024).
Wu, H. et al. AttentionMGT-DTA: A multi-modal drug-target affinity prediction using graph transformer and attention mechanism. Neural Netw. 169, 623–636 (2024).
Wang, J. et al. Predicting drug–target binding affinity with cross-scale graph contrastive learning. Brief. Bioinform. 25, bbad516 (2024).
Bi, X. et al. HiSIF-DTA: a hierarchical semantic information fusion framework for drug-target affinity prediction. IEEE J. Biomed. Health Inform. 29, 1579–1590 (2025).
Kumar, S., Teli, M. K. & Kim, M. GPCR-IPL score: multilevel featurization of GPCR–ligand interaction patterns and prediction of ligand functions from selectivity to biased activation. Brief. Bioinform. 25, bbae105 (2024).
Zeng, X. et al. MvGraphDTA: multi-view-based graph deep model for drug-target affinity prediction by introducing the graphs and line graphs. BMC Biol. 22, 182 (2024).
Zhao, L., Wang, H. & Shi, S. PocketDTA: an advanced multimodal architecture for enhanced prediction of drug− target affinity from 3D structural data of target binding pockets. Bioinformatics 40, btae594 (2024).
Li, M. et al. Contrastive meta-learning for drug-target binding affinity prediction. In Proceedings 2022 IEEE International Conference on Bioinformatics and Biomedicine (BIBM). (2022).
Lee, E. et al. MetaDTA: meta-learning-based drug-target binding affinity prediction. ICLR2022 Machine Learning for Drug Discovery. (2022).
Wang, Y. et al. ZeroBind: a protein-specific zero-shot predictor with subgraph matching for drug-target interactions. Nat. Commun. 14, 7861 (2023).
Vettoruzzo, A. et al. Advances and challenges in meta-learning: A technical review. IEEE Trans. Pattern Anal. Mach. Intell. 46, 4763–4779 (2024).
Xu, Y. et al. Task-aware meta-learning paradigm for universal structural damage segmentation using limited images. Eng. Struct. 284, 115917 (2023).
Kumar, R., Deleu, T. & Bengio, Y. The effect of diversity in meta-learning. Proc. AAAI Conf. Artif. Intell. 37, 8396–8404 (2023).
Zheng, H. et al. Learning from models beyond fine-tuning. Nat. Mach. Intell. 7, 6–17 (2025).
Zhang, L. et al. Style uncertainty based self-paced meta learning for generalizable person re-identification. IEEE Trans. Image Process. 32, 2107–2119 (2023).
Öztürk, H., Özgür, A. & Ozkirimli, E. DeepDTA: deep drug–target binding affinity prediction. Bioinformatics 34, i821–i829 (2018).
Wang, R. et al. Limaml: Personalization of deep recommender models via meta learning. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining 5882–5892 (2024).
Jose, S. T. & Simeone, O. An information-theoretic analysis of the impact of task similarity on meta-learning. In Proceedings 2021 IEEE International Symposium on Information Theory (ISIT) (2021).
Tran-Nguyen, V. K., Jacquemard, C. & Rognan, D. LIT-PCBA: an unbiased data set for machine learning and virtual screening. J. Chem. Inf. Model. 60, 4263–4273 (2020).
Çevik, F. et al. An assessment of metal pollution in surface sediments of Seyhan dam by using enrichment factor, geoaccumulation index and statistical analyses. Environ. Monit. Assess. 152, 309–317 (2009).
He, X., Folkman, L. & Borgwardt, K. Kernelized rank learning for personalized drug recommendation. Bioinformatics 34, 2808–2816 (2018).
Truchon, J. F. & Bayly, C. I. Evaluating virtual screening methods: Good and bad metrics for the “early recognition” problem. J. Chem. Inf. Model. 47, 488–508 (2007).
Zhou, G. et al. S-molsearch: 3d semi-supervised contrastive learning for bioactive molecule search. Adv. Neural Inf. Process. Syst. 37, 74715–74737 (2024).
Lan, J. et al. Structure-aware contrastive learning with fine-grained binding representations for drug discovery. Preprint at https://doi.org/10.48550/arXiv.2509.14788 (2025).
Morris, C. J. et al. MILCDock: machine learning enhanced consensus docking for virtual screening in drug discovery. J. Chem. Inf. Model. 62, 5342–5350 (2022).
Döhner, H. et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood 129, 424–447 (2017).
Wishart, D. S. et al. DrugBank 5.0: a major update to the DrugBank database for 2018. Nucleic Acids Res. 46, D1074–D1082 (2018).
Gu, S. et al. Benchmarking AI-powered docking methods from the perspective of virtual screening. Nat. Mach. Intell. 7, 509–520 (2025).
Kim, S. et al. PubChem 2023 update. Nucleic Acids Res. 51, D1373–D1380 (2023).
Stone, R. M. et al. Midostaurin plus chemotherapy for acute myeloid leukemia with a FLT3 mutation. New Engl. J. Med. 377, 454–464 (2017).
Ramaswamy, K. et al. Clofarabine with topotecan, vinorelbine, and thiotepa reinduction regimen for children and young adults with relapsed AML. Blood Adv. 6, 2688–2694 (2022).
Kivioja, J. L. et al. Dasatinib and navitoclax act synergistically to target NUP98-NSD1+/FLT3-ITD+ acute myeloid leukemia. Leukemia 33, 1360–1372 (2019).
Trott, O. & Olson, A. J. AutoDock Vina: Improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J. Comput. Chem. 31, 455–461 (2010).
Zegzouti, H., Zdanovskaia, M., Hsiao, K. & Goueli, S. A. ADP-Glo: A bioluminescent and homogeneous ADP monitoring assay for kinases. Assay Drug Dev. Technol. 7, 560–572 (2009).
Zarrinkar, P. P. et al. AC220 is a uniquely potent and selective inhibitor of FLT3 for the treatment of acute myeloid leukemia (AML). Blood 114, 2984–2992 (2009).
Sulzmaier, F. J., Jean, C. & Schlaepfer, D. D. FAK in cancer: mechanistic findings and clinical applications. Nat. Rev. Cancer 14, 598–610 (2014).
Meshinchi, S. et al. Clinical implications of FLT3 mutations in pediatric AML. Blood 108, 3654–3661 (2006).
Friedman, R. The molecular mechanisms behind activation of FLT3 in acute myeloid leukemia and resistance to therapy by selective inhibitors. Biochim. Biophys. Acta Rev. Cancer 1877, 188666 (2022).
Liu, T. et al. BindingDB: a web-accessible database of experimentally determined protein–ligand binding affinities. Nucleic Acids Res. 35, D198–D201 (2007).
Tang, J. et al. Making sense of large-scale kinase inhibitor bioactivity data sets: a comparative and integrative analysis. J. Chem. Inf. Model. 54, 735–743 (2014).
Davis, M. et al. Comprehensive analysis of kinase inhibitor selectivity. Nat. Biotechnol. 29, 1046–1051 (2011).
Thakur, A. et al. PIC50: An open source tool for interconversion of PIC50 values and IC50 for efficient data representation and analysis. Preprint at https://doi.org/10.1101/2022.10.15.512366 (2022).
Gil-Sorribes, M. Serrano, A. C. & Molina, A. Tensor-DTI: enhancing biomolecular interaction prediction with contrastive embedding learning. In Learning Meaningful Representations of Life (LMRL) Workshop at ICLR 2025 (2025).
Landrum, G. RDKit: A software suite for cheminformatics, computational chemistry, and predictive modeling. Greg Landrum 8, 5281 (2013).
Gharoun, H. et al. Meta-learning approaches for few-shot learning: A survey of recent advances. ACM Comput. Surv. 56, 1–41 (2024).
Finn, C., Abbeel, P. & Levine, S. Model-agnostic meta-learning for fast adaptation of deep networks. Int. Conf. Mach. Learn. 70, 1126–1135 (2017).
Scarselli, F. et al. The graph neural network model. IEEE Trans. Neural Netw. 20, 61–80 (2008).
Yao, H. et al. Meta-learning with an adaptive task scheduler. Adv. Neural Inf. Process. Syst. 34, 7497–7509 (2021).
Greff, K. et al. LSTM: A search space odyssey. IEEE Trans. Neural Netw. Learn. Syst. 28, 2222–2232 (2016).
Minot, M. & Reddy, S. T. Meta learning addresses noisy and under-labeled data in machine learning-guided antibody engineering. Cell Syst. 15, 4–18 (2024).
Hu, L. et al. ℓ-DARTS: Light-weight differentiable architecture search with robustness enhancement strategy. Knowledge Based Syst. 288, 111466 (2024).
Steck, H. et al. On ranking in survival analysis: Bounds on the concordance index. In Proceedings Advances in Neural Information Processing Systems 20 (2007).
Paszke, A. et al. PyTorch: An imperative style, high-performance deep learning library. In Proceedings Advances in Neural Information Processing Systems 32 (2019).
Gu, shukai. VSDS-VD: benchmarking AI-powered docking methods from the perspective of virtual screening. Zenodo. https://doi.org/10.5281/zenodo.13684010 (2024)
Wan, M. A meta learning and task adaptive approach for drug target affinity prediction. zenodo. https://doi.org/10.5281/zenodo.18595083.
Wallace, A. C., Laskowski, R. A. & Thornton, J. M. LIGPLOT: a program to generate schematic diagrams of protein-ligand interactions. Protein Eng. 8, 127–134 (1995).
Acknowledgements
This work was supported by the National Key R&D Program of China (Grant No. 2023YFC2604400, 2024YFA1307700), Shanghai Natural Science Foundation (Grant No. 25ZR1402171), the National Natural Science Foundation of China (62573425), PostGraduate Innovation Fund of Interdiscipline and New Medicine from School of Medicine of Shanghai University. Funding was provided to S.H. (2023YFC2604400, 62573425), X.B. (2024YFA1307700), and Y. Zhao (25ZR1402171).
Author information
Authors and Affiliations
Contributions
M.W., Y. Zhao, H.X., S.H. and X.B. conceived the study; M.W., Y. Zhang performed the experiments; M.W., Y. Zhao., H.X., and D.Y. conducted the surveys and collated the data; M.W., Y. Zhao, and Y. Zhang performed the writing-primer preparation; S.H. and X.B. performed the writing-reviewing and editing, and S.H., P.Z., and X.B. supervised the study. All authors have read and agreed to the published version of the manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks Xiaoyong Pan and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Wan, M., Zhao, Y., Zhang, Y. et al. A meta learning and task adaptive approach for drug target affinity prediction. Nat Commun (2026). https://doi.org/10.1038/s41467-026-70554-5
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-026-70554-5
