Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Article
  • Published:

Generalizable mutation-effect prediction across adaptive immune recognition via unified multimodal framework

Nature Machine Intelligence (2026) Cite this article

Subjects

Abstract

Adaptive immunity is a central defence system essential for long-term and highly specific protection against pathogens through the precise molecular recognition of antigens by lymphocytes. However, predicting how mutations reshape these interactions remains a major challenge. Although previous computational approaches leverage large-scale pretraining for mutation-effect predictions, most are designed for specific tasks or modalities and struggle to generalize across the heterogeneous, multimodal landscape of immune recognition. Here we introduce UniAIR, a modular, multimodal framework for the accurate and generalizable prediction of mutation effects across immune recognition scenarios. UniAIR integrates a standardized data pipeline, an interface-centric sequence–structure fusion transformer that integrates evolutionary information with geometric representations, and a suite of extensions for multiexpert consensus and adaptation to predicted structure inputs. We comprehensively evaluated UniAIR through large-scale benchmarking and independent tests across immunological tasks. The evaluation covered both extracellular and intracellular immune recognition, including antibody maturation, antigen escape, TCR–pHLA optimization and analyses in which experimental structures were unavailable. Extensive experiments show that UniAIR achieves state-of-the-art performance and delivers robust predictions with minimal task-specific tuning. In particular, UniAIR successfully performed multiround peptide optimization of a TCR–pHLA complex under sparse feedback and identified key functional mutations in incomplete antibody–antigen structures. Together, UniAIR establishes a unified computational foundation for mapping mutation landscapes, advancing understanding of adaptive immune recognition and accelerating immunotherapeutic design.

This is a preview of subscription content, access via your institution

Access options

Buy this article

  • Purchase on SpringerLink
  • Instant access to the full article PDF.

USD 39.95

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: UniAIR framework for mutation-effect prediction and immune-related applications.
The alternative text for this image may have been generated using AI.
Fig. 2: UniAIR achieves state-of-the-art performance on SKEMPIv2 and two independent test sets.
The alternative text for this image may have been generated using AI.
Fig. 3: UniAIR unveils the mutational patterns within immune-related complexes by zero-shot prediction.
The alternative text for this image may have been generated using AI.
Fig. 4: UniAIR uncovers antigen escape potential fine-tuned with few-shot annotations.
The alternative text for this image may have been generated using AI.
Fig. 5: UniAIR optimizes high-affinity peptide mutants by interacting with FEP.
The alternative text for this image may have been generated using AI.
Fig. 6: UniAIR can be extended to predict mutation effects with predicted structures.
The alternative text for this image may have been generated using AI.

Similar content being viewed by others

Data availability

The datasets used in this work are available at https://huggingface.co/datasets/Jesse7/UniAIR_data. The curated pretraining datasets are available at SabDab60 (https://opig.stats.ox.ac.uk/webapps/sabdab-sabpred/sabdab), STCRDab61 (https://opig.stats.ox.ac.uk/webapps/stcrdab-stcrpred) and PPB-Affinity62 (https://doi.org/10.1038/s41597-024-03997-4). The curated evaluation datasets are from SKEMPIv2 (ref. 63; https://life.bsc.es/pid/skempi2) and TCRen benchmark no. 1 (ref. 56; https://doi.org/10.1038/s43588-024-00653-0). For the curated downstream datasets, the HER2 test set is from the original study42 (https://doi.org/10.1101/2023.01.08.523187). The TCR–pMHC test set is from ATLAS43 (https://atlas.ibbr.umd.edu/web/index.php). The LASSA dataset is from ref. 47 (https://doi.org/10.1016/j.immuni.2024.06.013). The SARS-CoV-2 dataset is from ref. 5 (https://doi.org/10.1073/pnas.2122954119). The TCR–pHLA structures are from ref. 50 (https://doi.org/10.1038/s42256-024-00901-y). The KRAS complex is built from the RCSB PDB (https://www.rcsb.org/) with PDB ID 6ULR.

Code availability

The deep learning models were developed and deployed using standard model libraries and the PyTorch framework. The source code and model weights of UniAIR are available via GitHub at https://github.com/hanrthu/UniAIR and Zenodo at https://zenodo.org/records/19471285 (ref. 83).

References

  1. Rossjohn, J. et al. T cell antigen receptor recognition of antigen-presenting molecules. Annu. Rev. Immunol. 33, 169–200 (2015).

    Article  Google Scholar 

  2. Batista, F. D. & Harwood, N. E. The who, how and where of antigen presentation to B cells. Nat. Rev. Immunol. 9, 15–27 (2009).

    Article  Google Scholar 

  3. Paludan, S. R., Pradeu, T., Masters, S. L. & Mogensen, T. H. Constitutive immune mechanisms: mediators of host defence and immune regulation. Nat. Rev. Immunol. 21, 137–150 (2021).

    Article  Google Scholar 

  4. Flajnik, M. F. & Kasahara, M. Origin and evolution of the adaptive immune system: genetic events and selective pressures. Nat. Rev. Genet. 11, 47–59 (2010).

    Article  Google Scholar 

  5. Shan, S. et al. Deep learning guided optimization of human antibody against SARS-CoV-2 variants with broad neutralization. Proc. Natl Acad. Sci. USA 119, e2122954119 (2022).

    Article  Google Scholar 

  6. Pappas, L. et al. Rapid development of broadly influenza neutralizing antibodies through redundant mutations. Nature 516, 418–422 (2014).

    Article  Google Scholar 

  7. Reddehase, M. J. Antigens and immunoevasins: opponents in cytomegalovirus immune surveillance. Nat. Rev. Immunol. 2, 831–844 (2002).

    Article  Google Scholar 

  8. Vázquez-García, I. et al. Ovarian cancer mutational processes drive site-specific immune evasion. Nature 612, 778–786 (2022).

    Article  Google Scholar 

  9. Willmann, K. L. et al. Biallelic loss-of-function mutation in NIK causes a primary immunodeficiency with multifaceted aberrant lymphoid immunity. Nat. Commun. 5, 5360 (2014).

    Article  Google Scholar 

  10. Wang, G. et al. Deep-learning-enabled protein–protein interaction analysis for prediction of SARS-CoV-2 infectivity and variant evolution. Nat. Med. 29, 2007–2018 (2023).

    Article  Google Scholar 

  11. Wiehe, K. et al. Mutation-guided vaccine design: a process for developing boosting immunogens for HIV broadly neutralizing antibody induction. Cell Host Microbe 32, 693–709.e697 (2024).

    Article  Google Scholar 

  12. Cheng, J., Liang, T., Xie, X.-Q., Feng, Z. & Meng, L. A new era of antibody discovery: an in-depth review of AI-driven approaches. Drug Discov. Today 29, 103984 (2024).

    Article  Google Scholar 

  13. Cho, S. et al. Structural basis of affinity maturation and intramolecular cooperativity in a protein-protein interaction. Structure 13, 1775–1787 (2005).

    Article  Google Scholar 

  14. Homola, J. Surface plasmon resonance sensors for detection of chemical and biological species. Chem. Rev. 108, 462–493 (2008).

    Article  Google Scholar 

  15. Lequin, R. M. Enzyme immunoassay (EIA)/enzyme-linked immunosorbent assay (ELISA). Clin. Chem. 51, 2415–2418 (2005).

    Article  Google Scholar 

  16. Fowler, D. M., Stephany, J. J. & Fields, S. Measuring the activity of protein variants on a large scale using deep mutational scanning. Nat. Protoc. 9, 2267–2284 (2014).

    Article  Google Scholar 

  17. Frank, F. et al. Deep mutational scanning identifies SARS-CoV-2 Nucleocapsid escape mutations of currently available rapid antigen tests. Cell 185, 3603–3616.e3613 (2022).

    Article  Google Scholar 

  18. Boder, E. T. & Wittrup, K. D. Yeast surface display for screening combinatorial polypeptide libraries. Nat. Biotechnol. 15, 553–557 (1997).

    Article  Google Scholar 

  19. McMahon, C. et al. Yeast surface display platform for rapid discovery of conformationally selective nanobodies. Nat. Struct. Mol. Biol. 25, 289–296 (2018).

    Article  Google Scholar 

  20. Gerasimavicius, L., Livesey, B. J. & Marsh, J. A. Correspondence between functional scores from deep mutational scans and predicted effects on protein stability. Protein Sci. 32, e4688 (2023).

    Article  Google Scholar 

  21. Mason, D. M. et al. Optimization of therapeutic antibodies by predicting antigen specificity from antibody sequence via deep learning. Nat. Biomed. Eng. 5, 600–612 (2021).

    Article  Google Scholar 

  22. Brandes, N., Goldman, G., Wang, C. H., Ye, C. J. & Ntranos, V. Genome-wide prediction of disease variant effects with a deep protein language model. Nat. Genet. 55, 1512–1522 (2023).

    Article  Google Scholar 

  23. Hie, B., Zhong, E. D., Berger, B. & Bryson, B. Learning the language of viral evolution and escape. Science 371, 284–288 (2021).

    Article  MathSciNet  Google Scholar 

  24. Shanker, V. R., Bruun, T. U., Hie, B. L. & Kim, P. S. Unsupervised evolution of protein and antibody complexes with a structure-informed language model. Science 385, 46–53 (2024).

    Article  Google Scholar 

  25. van Beusekom, B. et al. Homology-based hydrogen bond information improves crystallographic structures in the PDB. Protein Sci. 27, 798–808 (2018).

    Article  Google Scholar 

  26. Waman, V. P. et al. CATH v4.4: major expansion of CATH by experimental and predicted structural data. Nucleic Acids Res. 53, D348–D355 (2024).

    Article  Google Scholar 

  27. McMaster, B., Thorpe, C., Ogg, G., Deane, C. M. & Koohy, H. Can AlphaFold’s breakthrough in protein structure help decode the fundamental principles of adaptive cellular immunity?. Nat. Methods 21, 766–776 (2024).

    Article  Google Scholar 

  28. Han, W. et al. Predicting the antigenic evolution of SARS-COV-2 with deep learning. Nat. Commun. 14, 3478 (2023).

    Article  Google Scholar 

  29. Taft, J. M. et al. Deep mutational learning predicts ACE2 binding and antibody escape to combinatorial mutations in the SARS-CoV-2 receptor-binding domain. Cell 185, 4008–4022.e4014 (2022).

    Article  Google Scholar 

  30. Hie, B. L. et al. Efficient evolution of human antibodies from general protein language models. Nat. Biotechnol. 42, 275–283 (2024).

    Article  Google Scholar 

  31. Cai, H. et al. Pretrainable geometric graph neural network for antibody affinity maturation. Nat. Commun. 15, 7785 (2024).

    Article  Google Scholar 

  32. Visani, G. M. et al. T-cell receptor specificity landscape revealed through de novo peptide design. Proc. Natl Acad. Sci. USA 122, e2504783122 (2025).

    Article  Google Scholar 

  33. O’Donnell, T. J. et al. Reading the repertoire: progress in adaptive immune receptor analysis using machine learning. Cell System 15, 1168–1189 (2024).

    Article  Google Scholar 

  34. Xie, L. et al. AI developments for T and B cell receptor modeling and therapeutic design. Preprint at https://arxiv.org/abs/2601.17138 (2026).

  35. Mhanna, V. et al. Adaptive immune receptor repertoire analysis. Nat. Rev. Methods Primers 4, 6 (2024).

  36. Lin, Z. et al. Evolutionary-scale prediction of atomic-level protein structure with a language model. Science 379, 1123–1130 (2023).

    Article  MathSciNet  Google Scholar 

  37. Guo, H. et al. Advancing expert specialization for better MoE. Adv. Neural Inf. Process. Syst. 38, 48767–48809 (2026).

    Google Scholar 

  38. Luo, S. et al. Rotamer density estimator is an unsupervised learner of the effect of mutations on protein–protein interaction. In The Eleventh International Conference on Learning Representations (2023).

  39. Bushuiev, A. et al. Learning to design protein-protein interactions with enhanced generalization. In International Conference on Learning Representations 21010–21035 (2025).

  40. Su, J. et al. SaProt: protein language modeling with structure-aware vocabulary. In International Conference on Learning Representations 6987–7009 (2024).

  41. Li, M. et al. ProSST: protein language modeling with quantized structure and disentangled attention. Adv. Neural Inf. Process. Syst. 37, 35700–35726 (2024).

  42. Shanehsazzadeh, A. et al. Unlocking de novo antibody design with generative artificial intelligence. Preprint at bioRxiv https://doi.org/10.1101/2023.01.08.523187 (2024).

  43. Borrman, T. et al. ATLAS: a database linking binding affinities with structures for wild-type and mutant TCR-pMHC complexes. Proteins Struct. Funct. Bioinf. 85, 908–916 (2017).

    Article  Google Scholar 

  44. Gao, Y. et al. Pan-peptide meta learning for T-cell receptor–antigen binding recognition. Nat. Mach. Intell. 5, 236–249 (2023).

    Article  Google Scholar 

  45. Sidney, J., Peters, B., Frahm, N., Brander, C. & Sette, A. HLA class I supertypes: a revised and updated classification. BMC Immunol. 9, 1 (2008).

    Article  Google Scholar 

  46. Garry, R. F. Lassa fever—the road ahead. Nat. Rev. Microbiol. 21, 87–96 (2023).

    Article  Google Scholar 

  47. Carr, C. R. et al. Deep mutational scanning reveals functional constraints and antibody-escape potential of Lassa virus glycoprotein complex. Immunity 57, 2061–2076.e2011 (2024).

    Article  Google Scholar 

  48. Zhang, Y. et al. Epitope-anchored contrastive transfer learning for paired CD8+ T cell receptor–antigen recognition. Nat. Mach. Intell. 6, 1344–1358 (2024).

    Article  Google Scholar 

  49. Sim, M. J. W. et al. High-affinity oligoclonal TCRs define effective adoptive T cell therapy targeting mutant KRAS-G12D. Proc. Natl Acad. Sci. USA 117, 12826–12835 (2020).

    Article  Google Scholar 

  50. Feng, Z. et al. Sliding-attention transformer neural architecture for predicting T cell receptor–antigen–human leucocyte antigen binding. Nat. Mach. Intell. 6, 1216–1230 (2024).

    Article  Google Scholar 

  51. Ullanat, V., Jing, B., Sledzieski, S. & Berger, B. Learning the language of protein-protein interactions. Nat. Commun. 17, 1199 (2025).

    Article  Google Scholar 

  52. Bran, A. M. et al. Augmenting large language models with chemistry tools. Nat. Mach. Intell. 6, 525–535 (2024).

    Article  Google Scholar 

  53. Liu, A. et al. DeepSeek-V3.2: pushing the frontier of open large language models. Preprint at https://arxiv.org/abs/2512.02556 (2025).

  54. OpenAI. Introducing GPT-5.2. https://openai.com/index/introducing-gpt-5-2/ (2025).

  55. Korpela, D. et al. EPIC-TRACE: predicting TCR binding to unseen epitopes using attention and contextualized embeddings. Bioinformatics 39, btad743 (2023).

    Article  Google Scholar 

  56. Karnaukhov, V. K. et al. Structure-based prediction of T cell receptor recognition of unseen epitopes using TCRen. Nat. Comput. Sci. 4, 510–521 (2024).

    Article  Google Scholar 

  57. Springer, I., Tickotsky, N. & Louzoun, Y. Contribution of T cell receptor alpha and beta CDR3, MHC typing, V and J genes to peptide binding prediction. Front. Immunol. 12, 664514 (2021).

    Article  Google Scholar 

  58. Facundo, D. & Batista, M. S. N. Affinity dependence of the B cell response to antigen: a threshold, a ceiling, and the importance of off-rate. Immunity 8, 751–759 (1998).

    Article  Google Scholar 

  59. Irving, M. et al. Interplay between T cell receptor binding kinetics and the level of cognate peptide presented by major histocompatibility complexes governs CD8+ T cell responsiveness. J. Biol. Chem. 287, 23068–23078 (2012).

    Article  Google Scholar 

  60. Dunbar, J. et al. SAbDab: the structural antibody database. Nucleic Acids Res. 42, D1140–1146 (2014).

    Article  Google Scholar 

  61. Leem, J., de Oliveira, S. H. P., Krawczyk, K. & Deane, C. M. STCRDab: the structural T-cell receptor database. Nucleic Acids Res. 46, D406–D412 (2018).

    Article  Google Scholar 

  62. Liu, H. et al. PPB-Affinity: protein-protein binding affinity dataset for AI-based protein drug discovery. Sci. Data 11, 1316 (2024).

    Article  Google Scholar 

  63. Jankauskaitė, J., Jiménez-García, B., Dapkūnas, J., Fernández-Recio, J. & Moal, I. H. SKEMPI 2.0: an updated benchmark of changes in protein–protein binding energy, kinetics and thermodynamics upon mutation. Bioinformatics 35, 462–469 (2018).

    Article  Google Scholar 

  64. Rodrigues, C. H. M., Myung, Y., Pires, D. E. V. & Ascher, D. B. mCSM-PPI2: predicting the effects of mutations on protein–protein interactions. Nucleic Acids Res. 47, W338–W344 (2019).

    Article  Google Scholar 

  65. Nahta, R. & Esteva, F. J. HER2 therapy: molecular mechanisms of trastuzumab resistance. Breast Cancer Res. 8, 215 (2006).

    Article  Google Scholar 

  66. Huang, X., Pearce, R. & Zhang, Y. EvoEF2: accurate and fast energy function for computational protein design. Bioinformatics 36, 1135–1142 (2020).

    Article  Google Scholar 

  67. Pearce, R., Huang, X., Setiawan, D. & Zhang, Y. EvoDesign: designing protein–protein binding interactions using evolutionary interface profiles in conjunction with an optimized physical energy function. J. Mol. Biol. 431, 2467–2476 (2019).

    Article  Google Scholar 

  68. Ahdritz, G. et al. OpenFold: retraining AlphaFold2 yields new insights into its learning mechanisms and capacity for generalization. Nat. Methods 21, 1514–1524 (2024).

    Article  Google Scholar 

  69. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).

    Article  Google Scholar 

  70. Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W. & Klein, M. L. Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 79, 926–935 (1983).

    Article  Google Scholar 

  71. Berendsen, H. J. C., van der Spoel, D. & van Drunen, R. GROMACS: a message-passing parallel molecular dynamics implementation. Comput. Phys. Commun. 91, 43–56 (1995).

    Article  Google Scholar 

  72. Abraham, M. J. et al. GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1–2, 19–25 (2015).

    Article  Google Scholar 

  73. MacKerell, A. D. et al. All-atom empirical potential for molecular modeling and dynamics studies of proteins. J. Phys. Chem. B 102, 3586–3616 (1998).

    Article  Google Scholar 

  74. Darden, T., York, D. & Pedersen, L. Particle mesh Ewald: an Nlog(N) method for Ewald sums in large systems. J. Chem. Phys. 98, 10089–10092 (1993).

    Article  Google Scholar 

  75. Essmann, U. et al. A smooth particle mesh Ewald method. J. Chem. Phys. 103, 8577–8593 (1995).

    Article  Google Scholar 

  76. Hess, B., Bekker, H., Berendsen, H. J. C. & Fraaije, J. G. E. M. LINCS: a linear constraint solver for molecular simulations. J. Comput. Chem. 18, 1463–1472 (1997).

    Article  Google Scholar 

  77. Giovanni Bussi, D. D. Michele Parrinello canonical sampling through velocity rescaling. J. Chem. Phys. 126, 014101 (2007).

    Article  Google Scholar 

  78. Parrinello, M. & Rahman, A. Polymorphic transitions in single crystals: a new molecular dynamics method. J. Appl. Phys. 52, 7182–7190 (1981).

    Article  Google Scholar 

  79. Jorgensen, W. L. Free energy calculations: a breakthrough for modeling organic chemistry in solution. Acc. Chem. Res. 22, 184–189 (1989).

    Article  Google Scholar 

  80. Deng, Y. & Roux, B. Calculation of standard binding free energies: aromatic molecules in the T4 lysozyme L99A mutant. J. Chem. Theory Comput. 2, 1255–1273 (2006).

    Article  Google Scholar 

  81. Boresch, S., Tettinger, F., Leitgeb, M. & Karplus, M. Absolute binding free energies: a quantitative approach for their calculation. J. Phys. Chem. B 107, 9535–9551 (2003).

    Article  Google Scholar 

  82. Mobley, D. L., Chodera, J. D. & Dill, K. A. On the use of orientational restraints and symmetry corrections in alchemical free energy calculations. J. Chem. Phys. 125, 084902 (2006).

    Article  Google Scholar 

  83. Han, R. hanrthu/UniAIR: Initial Release of UniAIR v1.0.0. Zenodo https://doi.org/10.5281/zenodo.19471285 (2026).

Download references

Acknowledgements

This study was supported by grants from the National Science Foundation of China (T2541010 to T.C.; T2522008 to G.W.; 82522048, 62501406 to X.L.), the National Key R&D Program of China (2024YFF1207100, 2024YFF1207103 to T.C.; 2025YFF0515300, 2025YFF0515301 to S.W. and G.W.), Shenzhen Medical Research Fund (E250200620, E250200622 and E250200623 to S.W. and G.W.), the National Health and Medical Research Council of Australia (APP1127948, APP1144652 and APP2036864 to J.S.), Australian Research Council (LP220200614 to J.S.), the Scientific Research Innovation Capability Support Project for Young Faculty (SRICSPYF-ZY2025015 to G.W.) and the Fundamental Research Funds for the Beijing University of Posts and Telecommunications (grant number 2025AI4S18 to G.W.). This work was also supported by the Beijing National Research Center for Information Science and Technology (BNRist) and the Major and Seed Inter-Disciplinary Research projects awarded by Monash University. This study was also funded by New Cornerstone Science Foundation through the XPLORER PRIZE. The funders had no roles in the study design, data collection and analysis, publication decisions or manuscript preparation.

Author information

Author notes

  1. These authors contributed equally: Rong Han, Yumeng Zhang, Xiaohong Liu, Lei Fu.

Authors and Affiliations

  1. Department of Computer Science and Technology, Tsinghua University, Beijing, China

    Rong Han, Peidong Zhang, Xuanzhong Chen & Ting Chen

  2. Monash Biomedicine Discovery Institute and Department of Biochemistry and Molecular Biology, Monash University, Melbourne, Victoria, Australia

    Yumeng Zhang, Tong Pan, Jing Xu, Xiaoyu Wang & Jiangning Song

  3. Institute of Medical Artificial Intelligence, South China Hospital, Medical School, Shenzhen University, Shenzhen, China

    Xiaohong Liu, Lei Fu & Song Wu

  4. College of Electronics and Information Engineering, Shenzhen University, Shenzhen, China

    Xiaohong Liu, Lei Fu & Song Wu

  5. South Australian immunoGENomics Cancer Institute, Faculty of Health and Medical Sciences, The University of Adelaide, Adelaide, South Australia, Australia

    Jing Xu

  6. School of Environment, Tsinghua University, Beijing, China

    Jiesi Lei

  7. State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing, China

    Wuyang Lan, Changwei Ji & Guangyu Wang

  8. School of Science and Engineering (SSE), The Chinese University of Hong Kong, Shenzhen, China

    Shuguang Cui

Authors
  1. Rong Han
    View author publications

    Search author on:PubMed Google Scholar

  2. Yumeng Zhang
    View author publications

    Search author on:PubMed Google Scholar

  3. Xiaohong Liu
    View author publications

    Search author on:PubMed Google Scholar

  4. Lei Fu
    View author publications

    Search author on:PubMed Google Scholar

  5. Tong Pan
    View author publications

    Search author on:PubMed Google Scholar

  6. Jing Xu
    View author publications

    Search author on:PubMed Google Scholar

  7. Xiaoyu Wang
    View author publications

    Search author on:PubMed Google Scholar

  8. Peidong Zhang
    View author publications

    Search author on:PubMed Google Scholar

  9. Xuanzhong Chen
    View author publications

    Search author on:PubMed Google Scholar

  10. Jiesi Lei
    View author publications

    Search author on:PubMed Google Scholar

  11. Wuyang Lan
    View author publications

    Search author on:PubMed Google Scholar

  12. Changwei Ji
    View author publications

    Search author on:PubMed Google Scholar

  13. Shuguang Cui
    View author publications

    Search author on:PubMed Google Scholar

  14. Song Wu
    View author publications

    Search author on:PubMed Google Scholar

  15. Jiangning Song
    View author publications

    Search author on:PubMed Google Scholar

  16. Ting Chen
    View author publications

    Search author on:PubMed Google Scholar

  17. Guangyu Wang
    View author publications

    Search author on:PubMed Google Scholar

Contributions

R.H., Y.Z., L.F., T.P., X.W., J.X., P.Z. and X.C. collected and analysed the data. R.H., J.X., X.W., T.P., W.L. and C.J. developed the models and downstream applications. S.C. provided high-performance computational resources and infrastructure support. G.W., T.C., J.S., S.W. and X.L. conceived of and supervised the project. R.H., Y.Z., G.W., X.L., J.S., T.C., T.P., J.X., X.W. and J.L. wrote and revised the paper. All authors discussed the results and reviewed the paper.

Corresponding authors

Correspondence to Song Wu, Jiangning Song, Ting Chen or Guangyu Wang.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Machine Intelligence thanks Miaozhe Huo, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Table 1 Detailed model comparison on SKEMPIv2
Full size table
Extended Data Table 2 Performance comparison of LLMs with tools on the SKEMPIv2 dataset
Full size table
Extended Data Table 3 Evaluation of UniAIR and TCR-pMHC binding prediction models in distinguishing cognate TCR epitopes from unrelated peptides
Full size table
Extended Data Table 4 Unseen evaluation of UniAIR and TCR-pMHC binding prediction models in distinguishing cognate TCR epitopes from unrelated peptides
Full size table

Supplementary information

Supplementary Information (download PDF )

Supplementary Figs. 1–7, Tables 1–6, baseline implementations and iDist embedding for dataset analysis.

Reporting Summary (download PDF )

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Han, R., Zhang, Y., Liu, X. et al. Generalizable mutation-effect prediction across adaptive immune recognition via unified multimodal framework. Nat Mach Intell (2026). https://doi.org/10.1038/s42256-026-01243-7

Download citation

  • Received:

  • Accepted:

  • Published:

  • Version of record:

  • DOI: https://doi.org/10.1038/s42256-026-01243-7

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing