基于虚拟科学家实现可编程功能性RNA的从头设计
Programmable functional RNA design with a virtual scientist
该研究在Nature Biotechnology发表,介绍了一种名为RhoAgent的AI驱动设计框架,用于可编程功能性RNA的从头设计。该方法结合扩散模型与虚拟科学家代理架构,可自主迭代完成RNA序列设计与功能验证,适用于核酶、适配体等多种功能性RNA元件的设计开发,相关代码已开源。
提出基于AI虚拟科学家的可编程功能性RNA设计框架,可生成目标结构的RNA分子序列及相关算法代码。
Data availability
The training dataset comprises experimentally determined RNA structures from the PDB (https://www.rcsb.org/), supplemented with sequences from the bpRNA database (https://bprna.cgrb.oregonstate.edu/) and the Ribonanza dataset (https://www.kaggle.com/competitions/stanford-ribonanza-rna-folding/data). All PDB structures used in this study are publicly available. Designed RNA sequences are provided in Supplementary Tables 2–20. Source data are provided with this paper.
Code availability
RhoAgent and RDiff are available via GitHub at https://github.com/ml4bio/RhoAgent and https://github.com/ml4bio/RDiff and via Zenodo at https://doi.org/10.5281/zenodo.22879438 (ref. 65). In addition, we used the software with the indicated version numbers in this study: cmbuild, cmcalibrate, cmscan and cmsearch from Infernal 1.1.3; cd-hit-est from CD-HIT 4.8.1; nhmmer from HMMER 3.3; HH-suite 2.0.15; Python = 3.8.12; NumPy = 1.21.2, a Python package for numerical calculations; PyTorch = 1.10.2, a deep learning framework in Python; pandas = 2.3.1, an open-source Python package for data analysis; r-base = 4.5.2; r-ggplot2 = 4.0.1, a system in R for declaratively creating graphics; SciPy = 1.14.1, an open-source Python package for mathematics; RNApolis is a Python library and utilities containing RNA-related bioinformatics functions and classes, available via GitHub at https://github.com/tzok/rnapolis-py; and Arena, a standalone program used for the full-atomic reconstruction of RNA structures from coarse-grained models, available via GitHub at https://github.com/pylelab/Arena.
References
Doudna, J. A. & Cech, T. R. The chemical repertoire of natural ribozymes. Nature 418, 222–228 (2002).
Mandal, M. & Breaker, R. R. Gene regulation by riboswitches. Nat. Rev. Mol. Cell Biol. 5, 451–463 (2004).
Warner, K. D., Hajdin, C. E. & Weeks, K. M. Principles for targeting RNA with drug-like small molecules. Nat. Rev. Drug Discov. 17, 547–558 (2018).
Kaseniit, K. E. et al. Modular, programmable RNA sensing using ADAR editing in living cells. Nat. Biotechnol. 41, 482–487 (2023).
Zhao, E. M. et al. RNA-responsive elements for eukaryotic translational control. Nat. Biotechnol. 40, 539–545 (2022).
Pardee, K. et al. Paper-based synthetic gene networks. Cell 159, 940–954 (2014).
Serganov, A. & Patel, D. J. Ribozymes, riboswitches and beyond: regulation of gene expression without proteins. Nat. Rev. Genet. 8, 776–790 (2007).
Iwano, N., Adachi, T., Aoki, K., Nakamura, Y. & Hamada, M. Generative aptamer discovery using RaptGen. Nat. Comput. Sci. 2, 378–386 (2022).
Zadeh, J. N. et al. NUPACK: analysis and design of nucleic acid systems. J. Comput. Chem. 32, 170–173 (2011).
Lorenz, R. et al. ViennaRNA Package 2.0. Algorithms Mol. Biol. 6, 26 (2011).
Riley, A. T., Robson, J. M., Ulanova, A. & Green, A. A. Generative and predictive neural networks for the design of functional RNA molecules. Nat. Commun. 16, 4155 (2025).
Sumi, S., Hamada, M. & Saito, H. Deep generative design of RNA family sequences. Nat. Methods 21, 435–443 (2024).
Behrouzi, R., Roh, J. H., Kilburn, D., Briber, R. M. & Woodson, S. A. Cooperative tertiary interaction network guides RNA folding. Cell 149, 348–357 (2012).
Golden, B. L., Kim, H. & Chase, E. Crystal structure of a phage Twort group I ribozyme–product complex. Nat. Struct. Mol. Biol. 12, 82–89 (2005).
Huang, L., Ishibe-Murakami, S., Patel, D. J. & Serganov, A. Long-range pseudoknot interactions dictate the regulatory response in the tetrahydrofolate riboswitch. Proc. Natl Acad. Sci. USA 108, 14801–14806 (2011).
Fu, Y., Sander, J. D., Reyon, D., Cascio, V. M. & Joung, J. K. Improving CRISPR-Cas nuclease specificity using truncated guide RNAs. Nat. Biotechnol. 32, 279–284 (2014).
Rockey, W. M. et al. Rational truncation of an RNA aptamer to prostate-specific membrane antigen using computational structural modeling. Nucleic Acid Ther. 21, 299–314 (2011).
Zhang, D. et al. Mutagenesis and structural studies reveal the basis for the specific binding of SARS-CoV-2 SL3 RNA element with human TIA1 protein. Nat. Commun. 14, 3715 (2023).
Ellington, A. D. & Szostak, J. W. In vitro selection of RNA molecules that bind specific ligands. Nature 346, 818–822 (1990).
Tuerk, C. & Gold, L. Systematic evolution of ligands by exponential enrichment: RNA ligands to bacteriophage T4 DNA polymerase. Science 249, 505–510 (1990).
Yesselman, J. D. et al. Computational design of three-dimensional RNA structure and function. Nat. Nanotechnol. 14, 866–873 (2019).
Wong, F. et al. Deep generative design of RNA aptamers using structural predictions. Nat. Comput. Sci. 4, 829–839 (2024).
Martick, M. & Scott, W. G. Tertiary contacts distant from the active site prime a ribozyme for catalysis. Cell 126, 309–320 (2006).
Ren, A. et al. In-line alignment and Mg2+ coordination at the cleavage site of the env22 twister ribozyme. Nat. Commun. 5, 5534 (2014).
Anderson-Lee, J. et al. Principles for predicting RNA secondary structure design difficulty. J. Mol. Bio. 428, 748–757 (2016).
Watson, J. L. et al. De novo design of protein structure and function with RFdiffusion. Nature 620, 1089–1100 (2023).
Ingraham, J. B. et al. Illuminating protein space with a programmable generative model. Nature 623, 1070–1078 (2023).
Ahern, W. et al. Atom-level enzyme active site scaffolding using RFdiffusion2. Nat. Methods 23, 96–105 (2026).
Das, R., Karanicolas, J. & Baker, D. Atomic accuracy in predicting and designing noncanonical RNA structure. Nat. Methods 7, 291–294 (2010).
Wang, J., Fan, Y., Hong, L., Hu, Z. & Li, Y. Deep learning for RNA structure prediction. Curr. Opin. Struct. Biol. 91, 102991 (2025).
Laing, C. & Schlick, T. Computational approaches to RNA structure prediction, analysis, and design. Curr. Opin. Struct. Biol. 21, 306–318 (2011).
Swanson, K., Wu, W., Bulaong, N. L., Pak, J. E. & Zou, J. The Virtual Lab of AI agents designs new SARS-CoV-2 nanobodies. Nature 646, 716–723 (2025).
Qu, Y. et al. CRISPR-GPT for agentic automation of gene-editing experiments. Nat. Biomed. Eng. 10, 245–258 (2026).
Xin, H., Kitchin, J. R. & Kulik, H. J. Towards agentic science for advancing scientific discovery. Nat. Mach. Intell. 7, 1373–1375 (2025).
Gottweis, J. et al. Accelerating scientific discovery with Co-Scientist. Nature 655, 487–496 https://doi.org/10.1038/s41586-026-10644-y (2026).
Gao, S. et al. Empowering biomedical discovery with AI agents. Cell 187, 6125–6151 (2024).
Ramos, M. C., Collison, C. J. & White, A. D. A review of large language models and autonomous agents in chemistry. Chem. Science 16, 2514–2572 (2025).
Chen, Y. et al. Hovlinc is a recently evolved class of ribozyme found in human lncRNA. Nat. Chem. Biol. 17, 601–607 (2021).
Jepsen, M. D. et al. Development of a genetically encodable FRET system using fluorescent RNA aptamers. Nat. Commun. 9, 18 (2018).
Abramson, J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature 630, 493–500 (2024).
Passaro, S. et al. Boltz-2: towards accurate and efficient binding affinity prediction. Preprint at bioRxiv https://doi.org/10.1101/2025.06.14.659707 (2025).
ByteDance AML AI4Science Team et al. Protenix-advancing structure prediction through a comprehensive AlphaFold3 reproduction. Preprint at bioRxiv https://doi.org/10.1101/2025.01.08.631967 (2025).
Shen, T. et al. Accurate RNA 3D structure prediction using a language model-based deep learning approach. Nat. Methods 21, 2287–2298 (2024).
Wang, W. et al. trRosettaRNA: automated prediction of RNA 3D structure with transformer network. Nat. Commun. 14, 7266 (2023).
Li, Y. et al. DRfold2 is a deep learning-based tool that enables efficient and accurate RNA structure prediction. PLoS Biol. 24, e3003659 (2026).
Lu, X.-J., Bussemaker, H. J. & Olson, W. K. DSSR: an integrated software tool for dissecting the spatial structure of RNA. Nucl. Acids Res. 43, e142 https://doi.org/10.1093/nar/gkv716 (2015).
Huang, K. et al. Autonomous biomedical research with an artificial intelligence agent. Science 393, eadz4351 (2026).
Yang, A. et al. Qwen3 technical report. Preprint at https://doi.org/10.48550/arXiv.2505.09388 (2025).
Sullenger, B. A. & Cech, T. R. Ribozyme-mediated repair of defective mRNA by targeted trans-splicing. Nature 371, 619–622 (1994).
Dykstra, P. B., Kaplan, M. & Smolke, C. D. Engineering synthetic RNA devices for cell control. Nat. Rev. Genet. 23, 215–228 (2022).
Quinodoz, S. A. et al. Mapping and engineering RNA-driven architecture of the multiphase nucleolus. Nature 644, 557–566 (2025).
Robertson, M. P. & Ellington, A. In vitro selection of an allosteric ribozyme that transduces analytes to amplicons. Nat. Biotechnol. 17, 62–66 (1999).
Tang, W., Hu, J. H. & Liu, D. R. Aptazyme-embedded guide RNAs enable ligand-responsive genome editing and transcriptional activation. Nat. Commun. 8, 15939 (2017).
Berman, H. M. et al. The protein data bank. Acta Crystallogr. Sect. D Biol. Crystallogr. 58, 899–907 (2002).
Adamczyk, B., Antczak, M. & Szachniuk, M. RNAsolo: a repository of cleaned PDB-derived RNA 3D structures. Bioinformatics 38, 3668–3670 (2022).
Krishna, R. et al. Generalized biomolecular modeling and design with RoseTTAFold All-Atom. Science 384, eadl2528 (2024).
Baek, M. et al. Accurate prediction of protein–nucleic acid complexes using RoseTTAFoldNA. Nat. Methods 21, 117–121 (2024).
Dai, T. et al. Autonomous mobile robots for exploratory synthetic chemistry. Nature 635, 890–897 (2024).
Szymanski, N. J. et al. An autonomous laboratory for the accelerated synthesis of inorganic materials. Nature 624, 86–91 (2023).
Lugmayr, A. et al. Repaint: inpainting using denoising diffusion probabilistic models. In Proc. IEEE/CVF Conference on Computer Vision and Pattern Recognition 11461–11471 (IEEE, 2022).
Danaee, P. et al. bpRNA: large-scale automated annotation and analysis of RNA secondary structure. Nucl. Acids Res. 46, 5381–5394 (2018).
He, S. et al. Ribonanza: deep learning of RNA structure through dual crowdsourcing. Preprint at bioRxiv https://doi.org/10.1101/2024.02.24.581671 (2024).
Steinegger, M. & Söding, J. MMseqs2 enables sensitive protein sequence searching for the analysis of massive data sets. Nat. Biotechnol. 35, 1026–1028 https://doi.org/10.1038/nbt.3988 (2017).
Ontiveros-Palacios, N. et al. Rfam 15: RNA families database in 2025. Nucl. Acids Res. 53, D258–D267 (2025).
Huang, H. et al. Supporting code for: programmable functional RNA design with a virtual scientist. Zenodo https://doi.org/10.5281/zenodo.22879438 (2026).
Acknowledgements
We would like to thank the Shenzhen Loop Area Institute for its support of this study.
Funding
This study was supported by the Shenzhen Medical Research Fund (award no. A2503002 to Y.L.), the National Key R&D Program of China (grant nos. 2025YFA0923500 to Y.L. and 2021YFA0909400 to D. Han), the National Natural Science Foundation of China (grant nos. 22225402 and 32341017 to D.Han and 32471343 and 32222089 to C.K.K.), the Postdoctoral Science Foundation of Zhejiang Province (grant no. ZJ2025004 to T.Z.), the Leading Innovative and Entrepreneur Team Introduction Program of Zhejiang Province (grant no. 2024R01005 to D. Han), the Chinese University of Hong Kong (CUHK; award nos. 4937025, 4937026, 5501517 and 5501329 to Y.L.), the IdeaBooster Fund (award nos. IDBF23ENG05 and IDBF24ENG06 to Y.L.), the Research Matching Grant Scheme at CUHK (award nos. 8601603 and 8601663 to Y.L.), the Research Grants Council of the Hong Kong Special Administrative Region (Hong Kong SAR), China (project nos. CUHK 24204023 and 14208525 to Y.L.; RFS2425-1S02, CityU 11100123, CityU 11100222 and CityU 11100421 to C.K.K.; PF22-73180 to J.W.), the Innovation and Technology Commission of the Hong Kong SAR, China (project nos. GHP/065/21SZ, ITS/247/23FP and PRP/033/24FX to Y.L.), the Croucher Foundation (project no. 9509003 to C.K.K.), the State Key Laboratory of Marine Pollution Seed Collaborative Research Fund (project nos. SCRF/0037, SCRF/0040 and SCRF/0070 to C.K.K.) and the City University of Hong Kong (project nos. 9680376, 7030001 and 9678302 to C.K.K.).
Author information
Author notes
These authors contributed equally: Han Huang, Tianqing Zhang, Liang Hong, Ziqian Lin, Dongchen He, Jiuming Wang.
Authors and Affiliations
Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong, Hong Kong SAR, China
Han Huang, Liang Hong, Ziqian Lin, Dongchen He, Jiuming Wang, Le Yu, Jiyue Jiang, Yimin Fan, Yanruisheng Shao, Irwin King, Pengan Chen & Yu Li
Hangzhou Institute of Medicine, Chinese Academy of Sciences, Hangzhou, China
Tianqing Zhang, Huimin Li & Da Han
Institute for Medical Engineering and Science, Massachusetts Institute of Technology, Cambridge, MA, USA
Tianqing Zhang & James J. Collins
Department of Chemistry and State Key Laboratory of Marine Environmental Health, City University of Hong Kong, Hong Kong, Hong Kong SAR, China
Tian-Ying Wu & Chun Kit Kwok
Hangzhou Institute for Advanced Study, University of Chinese Academy of Sciences, Hangzhou, China
Huimin Li
Division of Biological and Life Sciences, Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), Masdar City, United Arab Emirates
Yimin Fan & Pengan Chen
Shenzhen Research Institute of City University of Hong Kong, Shenzhen, China
Chun Kit Kwok
Wyss Institute for Biologically Inspired Engineering, Harvard University, Boston, MA, USA
James J. Collins
Broad Institute of MIT and Harvard, Cambridge, MA, USA
James J. Collins
Synthetic Biology Center, Massachusetts Institute of Technology, Cambridge, MA, USA
James J. Collins
The CUHK Shenzhen Research Institute, Shenzhen, China
Yu Li
Authors
- Han Huang
- Tianqing Zhang
- Liang Hong
- Ziqian Lin
- Dongchen He
- Jiuming Wang
- Tian-Ying Wu
- Huimin Li
- Da Han
- Le Yu
- Jiyue Jiang
- Yimin Fan
- Yanruisheng Shao
- Irwin King
- Pengan Chen
- Chun Kit Kwok
- James J. Collins
- Yu Li
Contributions
H.H., T.Z., L.H. and Y.L. conceived the study. H.H., L.H., Z.L. and D. He developed the computational model. H.H., L.H., D. He, L.Y. and Y.F. designed the agent framework. H.H., Z.L., J.W., J.J., Y.S., I.K. and P.C. processed the data and performed computational analyses. T.Z., T.-Y.W., H.L. and D. Han designed and performed the biological experiments. H.H., J.W., Z.L., L.H., D.He, C.K.K., J.J.C. and Y.L. wrote and edited the paper. T.Z., C.K.K., J.J.C. and Y.L. co-supervised the study.
Corresponding authors
Correspondence to Tianqing Zhang, Chun Kit Kwok, James J. Collins or Yu Li.
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Biotechnology thanks the anonymous reviewers for their contribution to the peer review of this work. Peer reviewer reports are available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data
Extended Data Fig. 1 SHAPE probing results of representative designs of CPEB3, twister, and hovlinc ribozymes using 2-methylnicotinic acid imidazolide (NAI).
a, Per-nucleotide SHAPE reactivities mapped onto the secondary structure input diagrams for three representative active designs, CP-05 (CPEB3, 67 nt), TW-1 (Twister, 73 nt), and HOV-13 (Hovlinc, 83 nt). Color indicates normalized NAI reactivity quantified from gel images. For visualization only, normalized reactivities greater than 1.2 were assigned to the maximum color category. b, Gel images from a SHAPE probing experiment for the three designs. Lane 1–4 in each gel show dideoxy sequencing (U, G, C, A), where each band identifies the corresponding nucleotide due to reverse transcription (RT) termination. Lane 5 (DMSO control) and 6 (NAI-modified) show SHAPE reactivity, where bands reflect NAI modification one nucleotide prior. Nucleotide identities, indices, and cleavage sites are annotated. Reactivities were derived by subtracting DMSO background from the NAI-modified signal and normalizing across the transcript.
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.
About this article
Cite this article
Huang, H., Zhang, T., Hong, L. et al. Programmable functional RNA design with a virtual scientist. Nat Biotechnol (2026). https://doi.org/10.1038/s41587-026-03362-3
Received: 22 January 2026
Accepted: 24 September 2026
Published: 19 November 2026
Version of record: 19 November 2026
DOI: https://doi.org/10.1038/s41587-026-03362-3
来源:Nature Biotechnology · nature.com