AI can speed up antibody design to block new virus: study

Artificial intelligence (AI) and “protein language” models can speed the design of monoclonal antibodies that can prevent or reduce the severity of potentially life-threatening viral infections, according to a multicenter study by researchers at Vanderbilt University Medical Center.

Their report was published Nov. 4 in the journal cellfocuses on developing antibody therapeutics against existing and emerging viral threats, including respiratory syncytial virus (RSV) and avian influenza virus, and the study’s implications are much broader, said the paper’s corresponding author. Ibelin GeorgievPh.D.

“This study is an important early milestone toward the ultimate goal of using computers to efficiently and effectively design new biologics from scratch and bring them to the clinic,” said Georgiev, professor of pathology, microbiology, and immunology and director of Vanderbilt University. program in computational microbiology and immunology.

“Such an approach would have a major positive impact on public health and could be applied to a wide range of diseases, including cancer, autoimmune, neurological diseases and many others,” he said.

Georgiev is a leader in the use of computational approaches to advance disease treatment and prevention. He is the principal investigator receiving an award of up to $30 million from the Advanced Research Projects Agency for Health (ARPA-H) to support the application of AI technologies that can develop novel antibodies with therapeutic potential.

Perry WadinA data scientist in Georgiev’s lab, he was involved in all aspects of the research and is the first author on the paper.

A research team of scientists from across Australia and Sweden has shown that protein language models can design functional human antibodies that recognize the unique antigenic sequences (surface proteins) of a particular virus, without requiring part of the antibody sequence as a starting template.

Protein language models are a type of large-scale language model (LLM) that are trained on vast amounts of text to enable language processing and generation. LLM provides the core functionality of chatbots such as ChatGPT.

By training the protein language model MAGE (monoclonal antibody generator) with previously characterized antibodies against known strains of the H5N1 influenza (avian influenza) virus, researchers were able to generate antibodies against a related but as yet unidentified strain of influenza.

These findings suggest that MAGE “may be used to rapidly generate antibodies against emerging health threats,” the researchers concluded, rather than traditional antibody discovery methods that require blood samples from infected people or antigenic proteins from new viruses.

Other co-authors of the Vanderbilt paper include Dr. Alexis Janke, Dr. Thoma Marinoff, Dr. Gwen Jordan, Dr. Olivia Powers, Dr. Matthew Vukovich, Dr. Clinton Holt and Dr. Alexandra Abu-Shmais.

This research was funded in part by the Advanced Research Projects Agency for Health (ARPA-H) and National Institutes of Health grants R01AI175245, R01AI152693, and 1ZIAAI005003. The views and conclusions contained in this document are those of the authors and should not be construed as representing official policy, express or implied, of the U.S. government.

Magician structure diagram: Cryo-electron microscopy (cryo-EM) resolution of the structure of a respiratory syncytial virus fusion protein (shades of pink) bound to fragments of two antibodies (dark/light and blue/green) designed by the researchers’ protein language model MAGE. Wasdin et al., Generation of antigen-specific paired-chain antibodies using large-scale language models, Cell (2025), © 2025 The Authors. Published by Elsevier https://doi.org/10.1016/j.cell.2025.10.006.