SEOUL, April 09 (AJP) - Researchers in South Korea have successfully used artificial intelligence to design proteins that can detect specific chemicals, including stress hormones, through a collaboration project between the Korea Advanced Institute of Science and Technology and 2024 Nobel laureate David Baker from the University of Washington, the prominent South Korean science institute said Thursday.
The study demonstrates a way to create functional biosensors from scratch, a process known as de novo design. Unlike previous methods that relied on modifying proteins already found in nature, this artificial intelligence-driven approach allows scientists to build custom proteins for specific tasks, such as medical testing or environmental monitoring.
Specifically, the Korea Advanced Institute of Science and Technology (KAIST) team created a protein that selectively identifies cortisol, a hormone closely linked to stress levels. By turning this protein into a working sensor, the researchers addressed a long-standing technical hurdle: designing proteins that can recognize small, complex molecules with atomic-level accuracy.
The technology could eventually lead to new tools for early disease diagnosis by spotting biomarkers in blood samples. It also has potential applications in the real-time monitoring of air and water quality or the development of more targeted medical treatments.
To design the proteins, the team developed an AI model that accounts for how amino acids interact with different chemicals. They tested the model by designing six different types of proteins for various drugs and metabolites, verifying the results through laboratory experiments and crystal structure analysis.
Lead researcher Lee Kyu-ri joined the KAIST Department of Biological Sciences in February 2025 after working as a staff scientist in the Baker laboratory in the United States. She now leads her own protein design lab, focusing on creating artificial enzymes, sensors, and RNA-recognition proteins.
The research was highlighted during a visit by David Baker to the KAIST campus on Thursday. "This research experimentally proves that we can use AI to design proteins that precisely recognize specific compounds," Lee said. "We plan to expand this technology to various fields such as disease diagnosis and the development of new drugs."
The project was supported by the Ministry of Science and ICT (MSIT) through the InnoCORE program. This initiative is designed to help South Korean researchers collaborate with global experts on advanced AI technologies.
KAIST President Lee Kwang-hyung noted that the collaboration with David Baker resulted in a significant achievement for AI-based protein design. The university currently manages several research units focused on aerospace, manufacturing, and drug discovery under the InnoCORE program.
(Reference Information)
Journal/Source: Nature Communications
Title: Small-molecule binding and sensing with a designed protein family
Link/DOI: https://bit.ly/4srcGAi
Copyright ⓒ Aju Press All rights reserved.