The Virtual Lab Of Ai Agents Designs New Sarscov2 Nanobodies Kyle Swanson by Kyle Swanson, Wesley Wu, Nash L. Bulaong, John E. Pak, James Zou 10.1038/S41586-025-09442-9 instant download after payment.

Abstract10 Science frequently benefits from teams of interdisciplinary researchers1–3 , but many 11 scientists do not have easy access to experts from multiple fields4,5 . While large language 12 models (LLMs) have shown an impressive ability to aid researchers across diverse 13 domains, their uses have been largely limited to answering specific scientific questions 14 rather than performing open-ended research6–11 . Here, we expand the capabilities of LLMs 15 for science by introducing the Virtual Lab, an AI-human research collaboration to perform 16 sophisticated, interdisciplinary science research. The Virtual Lab consists of an LLM 17 principal investigator agent guiding a team of LLM scientist agents through a series of 18 research meetings, with a human researcher providing high-level feedback. We apply the 19 Virtual Lab to design nanobody binders to recent variants of SARS-CoV-2. The Virtual 20 Lab creates a novel computational nanobody design pipeline that incorporates ESM, 21 AlphaFold-Multimer, and Rosetta and designs 92 new nanobodies. Experimental validation 22 reveals a range of functional nanobodies with promising binding profiles across SARS23 CoV-2 variants. In particular, two new nanobodies exhibit improved binding to the recent 24 JN.1 or KP.3 variants 12,13 while maintaining strong binding to the ancestral viral spike 25 protein, suggesting exciting candidates for further investigation. This demonstrates how the 26 Virtual Lab can rapidly make an impactful, real-world scientific discovery.