Skip to content
DraconiticResearch
Search Sources API
197,787 studies31 sourcesUpdated 35 min ago
Peer-reviewedCC BYNational Science Review

Predicting the evolutionary and functional landscapes of viruses with a unified nucleotide-protein language model: LucaVirus

  • Yuanfei Pan1,2,3
  • Yong He4
  • Yu-Qi Liu5,6,7,8
  • Yongtao Shan9,8
  • Shuning Liu10
  • Jia-Hao Ma1,2
  • Xue Liu1
  • Xiaoyun Pan3
  • Yinqi Bai11
  • Zan Xu4
  • Tingjun Hou12
  • Zheng Wang4
7 more authors
  • Jieping Ye4
  • Jianguo He13,9
  • Edward Charles Holmes14
  • Bo Li15,3
  • Yao-Qing Chen10
  • Zhaorong Li4,8
  • Mǎng Shī16,1,2
16 affiliations
  1. 1National Key Laboratory of Intelligent Tracking and Forecasting for Infectious Diseases, School of Medicine, Shenzhen Campus of Sun Yat-sen University, Sun Yat-sen University , Shenzhen 518107 ,
  2. 2State Key Laboratory for Biocontrol, School of Medicine, Shenzhen Campus of Sun Yat-sen University, Sun Yat-sen University , Shenzhen 518107 ,
  3. 3State Key Laboratory of Wetland Conservation and Restoration, National Observations and Research Station for Wetland Ecosystems of the Yangtze Estuary, Ministry of Education Key Laboratory for Biodiversity Science and Ecological Engineering, and Institute of Eco-Chongming, School of Life Sciences, Fudan University , Shanghai 200433 ,
  4. 4Apsara Lab, Alibaba Cloud Intelligence, Alibaba Group , Hangzhou 310030 ,
  5. 5BGI Research , Hangzhou 310030 ,
  6. 6HIM-BGI Omics Center, Hangzhou Institute of Medicine (HIM), Chinese Academy of Sciences , Hangzhou 310000 ,
  7. 7Hangzhou Institute for Advanced Study, University of Chinese Academy of Sciences , Hangzhou 310024 ,
  8. 8Zhejiang Lab , Hangzhou 311121 ,
  9. 9State Key Laboratory of Biocontrol, School of Life Science, Sun Yat-sen University , Guangzhou 510275 ,
  10. 10School of Public Health (Shenzhen), Shenzhen Campus of Sun Yat-sen University, Sun Yat-sen University , Shenzhen 518107 ,
  11. 11BGI Research , Sanya 572025 ,
  12. 12College of Pharmaceutical Sciences, Zhejiang University , Hangzhou 310058, Zhejiang ,
  13. 13School of Marine Sciences, Sun Yat-sen University State Key Laboratory of Biocontrol/Southern Laboratory of Ocean Science and Engineering (Guangdong, Zhuhai ), ; Guangzhou 510275 ,
  14. 14School of Medical Sciences, The University of Sydney , Sydney, NSW 2006 ,
  15. 15State Key Laboratory for Vegetation Structure, Functions and Construction (VegLab), Ministry of Education Key Laboratory for Transboundary Ecosecurity of Southwest China, Institute of Biodiversity, School of Ecology and Environmental Science, and Southwest United Graduate School, Yunnan University , Kunming 650500 ,
  16. 16Guangdong Provincial Center for Disease Control and Prevention , Guangzhou 511430 ,
Open at sourceDOICite

Abstract

Predicting viral evolution and function remains a central challenge in biology, hindered by high sequence divergence and limited knowledge compared to cellular organisms. Here, we introduce LucaVirus, a multi-modal foundation model for viruses, trained on 25.4 billion nucleotide and amino acid tokens covering a vast majority of catalogued viral diversity. LucaVirus learns biologically meaningful representations that reflect relationships between sequences, protein/gene homology, and evolutionary divergence. Using these embeddings, we developed downstream models that address key virology tasks: identifying hidden viruses in genomic 'dark matter', annotating enzymatic activities of uncharacterized proteins, predicting viral evolvability, and identifying antibody candidates for emerging viruses. LucaVirus demonstrates competitive performance in three tasks and matches leading models in the fourth with one-third the parameters. Together, these findings demonstrate the utility of a unified foundation model in analyzing viral sequence data and establish LucaVirus as an efficient and versatile platform for AI-driven virology, from virus discovery to functional and therapeutic predictions.

Related

  1. Non-adaptive evolutionary forces predominantly shape nucleotide composition near human transcription start sites
    Yi Qiu, Emilie Hebraud, Fanny Pouyet +1
    PLOS BiologyPeer-reviewedCC BY 4.0Full textMolecular biology & gene regulationGenetics & genomicsPLOSPMC
  2. Actuator Transcription Factors Resolve Combinatorial Binding Into Productive Gene Expression : An Evolutionary and Functional Framework for Gene Activation by Combinatorial TF Assemblies
    Nicoletta Bobola, Zoulfia Darieva
    BioEssays : news and reviews in molecular, cellular and developmental biologyPeer-reviewedCC BYFull textMolecular biology & gene regulationEurope PMC
  3. Predicting Functional Diversity Within Enzyme Families: A Practical Perspective
    Nika Sokolova, Kristina Haslinger
    Chembiochem : a European journal of chemical biologyPeer-reviewedCC BYFull textMolecular biology & gene regulationEurope PMC
  4. Dynamic control of nucleotide metabolism in physiology and disease
    Dohun Kim, Gerta Hoxhaj
    Nature cell biologyPeer-reviewedLicence not statedFull text1 citedMolecular biology & gene regulationPMC
  5. A fluorescent non-hydrolyzable probe for the nucleotide binding sites of KATP
    Prisicilla Rubio, Shayla Q. Whitaker, Jonathon Ashby +1
    The Journal of general physiologyPeer-reviewedLicence not statedFull textMolecular biology & gene regulationPMC

Cite

BibTeXRIS
BibTeX
@article{pan2026predicting,
  title = {{Predicting the evolutionary and functional landscapes of viruses with a unified nucleotide-protein language model: LucaVirus}},
  author = {Yuanfei Pan and Yong He and Yu-Qi Liu and Yongtao Shan and Shuning Liu and Jia-Hao Ma and Xue Liu and Xiaoyun Pan and Yinqi Bai and Zan Xu and Tingjun Hou and Zheng Wang and Jieping Ye and Jianguo He and Edward Charles Holmes and Bo Li and Yao-Qing Chen and Zhaorong Li and Mǎng Shī},
  journal = {National Science Review},
  year = {2026},
  month = jun,
  doi = {10.1093/nsr/nwag376},
  url = {https://doi.org/10.1093/nsr/nwag376},
  note = {CC BY},
}
RIS
TY  - JOUR
TI  - Predicting the evolutionary and functional landscapes of viruses with a unified nucleotide-protein language model: LucaVirus
AU  - Yuanfei Pan
AU  - Yong He
AU  - Yu-Qi Liu
AU  - Yongtao Shan
AU  - Shuning Liu
AU  - Jia-Hao Ma
AU  - Xue Liu
AU  - Xiaoyun Pan
AU  - Yinqi Bai
AU  - Zan Xu
AU  - Tingjun Hou
AU  - Zheng Wang
AU  - Jieping Ye
AU  - Jianguo He
AU  - Edward Charles Holmes
AU  - Bo Li
AU  - Yao-Qing Chen
AU  - Zhaorong Li
AU  - Mǎng Shī
PY  - 2026
DA  - 2026/06/16
JO  - National Science Review
DO  - 10.1093/nsr/nwag376
UR  - https://doi.org/10.1093/nsr/nwag376
AB  - Predicting viral evolution and function remains a central challenge in biology, hindered by high sequence divergence and limited knowledge compared to cellular organisms. Here, we introduce LucaVirus, a multi-modal foundation model for viruses, trained on 25.4 billion nucleotide and amino acid tokens covering a vast majority of catalogued viral diversity. LucaVirus learns biologically meaningful representations that reflect relationships between sequences, protein/gene homology, and evolutionary divergence. Using these embeddings, we developed downstream models that address key virology tasks: identifying hidden viruses in genomic 'dark matter', annotating enzymatic activities of uncharacterized proteins, predicting viral evolvability, and identifying antibody candidates for emerging viruses. LucaVirus demonstrates competitive performance in three tasks and matches leading models in the fourth with one-third the parameters. Together, these findings demonstrate the utility of a unified foundation model in analyzing viral sequence data and establish LucaVirus as an efficient and versatile platform for AI-driven virology, from virus discovery to functional and therapeutic predictions.
SN  - 2053-714X
SN  - 2095-5138
ER  -