Professor Sang Yup Lee’s team at KAIST has presented groundbreaking research on designing enzymes using AI. (Image courtesy of KAIST)
DAEJEON, South Korea, April 19 — A research team at the Korea Advanced Institute of Science and Technology (KAIST) has proposed a groundbreaking approach using artificial intelligence to design enzymes that do not exist in nature — a move that could transform the future of bioengineering and industrial biotechnology.
According to KAIST’s announcement on April 17, 2025, Distinguished Professor Sang Yup Lee of the Department of Biological and Chemical Engineering led the study, which outlines the potential of advanced AI models to not only predict enzyme function but also design novel enzymes with entirely new capabilities.
Enzymes — proteins that catalyze biochemical reactions within cells — are critical to building microbial cell factories, where genetically modified microorganisms are used to mass-produce valuable compounds. Designing new enzymes with specific functions is a key challenge in this field.
The team reviewed a wide range of AI methodologies already applied to enzyme prediction, including convolutional neural networks, recurrent neural networks, graph neural networks, and transformer-based large language models.
Their analysis emphasized how these deep learning systems can extract meaningful information from amino acid sequences to enhance prediction accuracy.
Beyond traditional sequence similarity analysis, the researchers found that deep learning could automatically identify structural and evolutionary features embedded in protein sequences — crucial clues for understanding catalytic function.
Importantly, the study points to generative AI models as a future pathway for not only predicting existing enzyme functions but also creating entirely new enzymes with synthetic capabilities absent from the natural world.
“AI-powered enzyme prediction and design will dramatically reshape the trajectory of biotechnology and life science research,” Professor Lee stated.
The findings were published in the March 28, 2025 edition of the international journal Trends in Biotechnology.
Kevin Lee (kevinlee@koreabizwire.com)
