The approach aims to rapidly generate antibodies against emerging threats, potentially outpacing conventional discovery methods that rely on patient samples or isolated viral proteins.

The research, published in Cell on November 4, 2025, indicates that AI-driven protein language models can accelerate monoclonal antibody design to fight viral infections, including RSV and avian influenza.

Collaborators from the United States, Australia, and Sweden contributed to the work, with funding support from ARPA-H and NIH grants R01AI175245, R01AI152693, and 1ZIAAI005003.

Corresponding author Ivelin Georgiev highlights the broader potential of computational design for therapeutics across diseases, aiming to translate designed biologics into the clinic.

The study suggests wider applicability beyond influenza and RSV, envisioning future use in cancer, autoimmunity, neurological diseases, and other conditions, as a step toward designing novel biologics from scratch with computational methods.

Computational design is positioned as a complement or potential replacement for traditional antibody discovery, which often depends on patient samples or antigens from new viruses.

Funding support for the project comes from ARPA-H and NIH (R01AI175245, R01AI152693, 1ZIAAI005003).

A Vanderbilt-led study demonstrates that AI-powered protein language models, specifically the MAGE system trained on known antibodies against H5N1, can design antibodies targeting related but unseen influenza strains, showing recognition of virus-specific antigen sequences without starting templates.

The Cell publication on November 4 underscores that AI-based antibody design could accelerate responses to emerging health threats relative to traditional approaches.

Perry Wasdin, the Georgiev lab data scientist, is listed as the paper’s first author and helped contribute to all aspects of the study, which involved Vanderbilt and international collaborators.

The paper, titled Generation of antigen-specific paired-chain antibodies using large language models, includes cryo-EM visuals of antibody-antigen complexes designed by MAGE and was published in Cell on 4-Nov-2025.

Georgiev emphasizes the long-term goal of bringing AI-designed biologics into clinical use to bolster public health.