"Using artificial intelligence technology, we can design and synthesize proteins that do not exist in nature. Generative AI is now the core technology for next-generation protein design."
A Nobel Prize-winning scholar who laid the foundation for "nanomedicine" through AI-powered protein design technology predicted that "generative AI will fundamentally transform the structure of entire industries — from next-generation infectious disease treatment and new drug development to chemical processes and recycling." The assessment is that the convergence of biomedical engineering and AI technology has opened an era in which nanometer-scale (one-billionth of a meter) proteins can be precisely designed to perform desired functions.
On Tuesday, David Baker, a professor at the University of Washington, made these remarks during a keynote speech at the "Yonsei-IBS Nobel Forum," jointly hosted by the Yonsei Institute for Advanced Science (YIAS) and the Institute for Basic Science (IBS) Center for Nanomedicine. This marks Baker's first visit to South Korea since winning the 2024 Nobel Prize in Chemistry.
Baker received the 2024 Nobel Prize in Chemistry for achievements including the development of "RoseTTAFold Diffusion (RF diffusion)," an AI model for generating protein structures. RF is an AI model Baker developed, inspired by AlphaFold, that decodes and designs protein structures. He has recently focused on developing machine learning techniques to generate functional proteins as follow-up research.
Protein design represents an approach that reverses the flow of traditional biology research. Previously, researchers started from genetic information, analyzed amino acid sequences and protein structures, and then understood their functions. Now, they start from the goal of "what function should the protein perform" and reverse-engineer the structure and sequence accordingly. "We can now reproduce and even expand the roles proteins have performed in nature," Baker said. "By repeatedly training the RF model to remove noise from protein structures and assigning conditions so it performs only specific functions such as binding to cancer cells, we have been able to create entirely new protein structures."
Baker is verifying the applicability of protein nanotechnology in the treatment of infectious diseases, immune disorders, and cancer. For example, specific proteins are designed to selectively react with and bind only to cancer cells or viruses, performing removal functions and dramatically reducing side effects during treatment. "It is possible to develop customized drugs for individual patients based on cancer antigens that differ from person to person," he said. "We have confirmed multiple times that AI-designed proteins effectively bind to targeted disordered proteins within cells."
This technology could also selectively remove "tau" proteins, identified as a causative substance of Alzheimer's disease. Baker introduced an "AI-designed protease (protein-degrading enzyme)" that recognizes and cleaves or induces degradation of only abnormally accumulated phosphorylated tau proteins in the brain. "This is a quintessential example of a precision nanomachine that could be used to treat neurodegenerative diseases," he said.
Baker also explained that even if a global pandemic like COVID-19 were to occur again, "nanobodies" targeting only specific pathogens could be designed for faster response. Nanobodies, the smallest form of single-domain antibodies, are much smaller than conventional antibodies and can penetrate deep into viruses to neutralize them.
Baker predicted that AI-based protein design technology has significant potential to spread beyond medicine into entire industries including chemical processes. "Currently, most industrial processes rely on toxic chemicals, but newly designed enzymes could replace them," he said. "More environmentally friendly processes will become possible." He added that nano-protein technology can also be applied to plastic degradation and photosynthesis efficiency improvement, noting that related research is already producing results.
Baker stressed the need for investment to enhance versatility, saying "protein design stands at a very important turning point right now." He said, "The range of applications will continue to expand." However, he pointed out that "the medical field has a well-established investment ecosystem where promising technologies can quickly spin off into startups and attract funding, whereas fields like the chemical industry lack sufficient investment." He added, "We already have the capability to design nanotechnology beyond medicine, but there are difficulties in translating this into reality."
The forum was also attended by Randy Schekman, the 2013 Nobel Prize laureate in Physiology or Medicine; Hannele Ruohola-Baker, a professor at the University of Washington; Itai Cohen, a professor at Cornell University; Cheon Jin-woo, dean of the Yonsei Institute for Advanced Science and director of the IBS Center for Nanomedicine; and Han Nam-sik, a professor at Yonsei University and director of the AI Research Lab at the Milner Therapeutics Institute at the University of Cambridge. They discussed the future direction of convergence research in AI, robotics, and nanomedicine.