A domestic research team, together with the lab of David Baker, the 2024 Nobel Prize in chemistry laureate and a professor at the University of Washington, used artificial intelligence (AI) to design artificial proteins that precisely recognize specific compounds and implemented them as actual Biosensors.
KAIST said on the 9th that the AI-CRED Innovative New Drug InnoCore Research Group, which includes Department of Biological Sciences Professor Lee Gyuri, succeeded through joint research with Professor Baker in using AI to design proteins that selectively recognize desired compounds. The results were published on Mar. 28 (local time) in the international journal Nature Communications.
Lee conducted research as a postdoctoral researcher and staff scientist at the University of Washington and the Howard Hughes Medical Institute (HHMI) from 2018 to 2024. Baker is a world-renowned authority in protein structure prediction and design.
The core of this study goes a step beyond the conventional approach of finding proteins in nature to use or improving only some functions, by designing de novo proteins with desired functions from the start and verifying whether they actually work. The team custom-built proteins that respond to specific compounds through AI-based design and even succeeded in linking them to sensor technology.
In particular, after designing a protein that selectively recognizes the stress hormone cortisol, the team implemented a Biosensor capable of actual measurement based on it. In the field of protein design, technology to accurately recognize small-molecule compounds has long been considered a representative challenge.
The researchers developed an AI model that can precisely capture protein–ligand interactions and, based on it, began designing new binding proteins. As a result, they designed artificial binding proteins for each of six compounds, including metabolites and small-molecule drugs, and confirmed their function through experiments.
Among them, using a new protein that binds cortisol, the team designed a chemical-induced dimer and succeeded in developing a cortisol Biosensor. The related design technology has been filed for a provisional patent in the United States.
Professor Lee Gyuri said, "This study is an experimental proof that AI can be used to design proteins that precisely recognize specific compounds," adding, "We will expand it into a protein design technology applicable to various fields, including disease diagnosis, drug development, and environmental monitoring."
References
Nature Communications (2026), DOI: https://doi.org/10.1038/s41467-026-70953-8