Australian scientists have published groundbreaking research in Nature Communications, successfully designing novel proteins using artificial intelligence for the first time that can effectively kill antibiotic-resistant bacteria. The study, jointly conducted by the Bio21 Institute at the University of Melbourne and the Monash Biomedicine Discovery Institute, reduced the protein development cycle from decades using traditional methods to just seconds.

The research team utilized an open-source AI protein design platform, integrating multiple tools including Bindcraft and Chai, to achieve rapid, customized protein design. Co-lead investigator Associate Professor Gavin Knott stated: "These proteins have applications across drugs, vaccines, and nanomaterials." The platform enables the design of proteins that specifically bind to target molecules, functioning as inhibitors or engineered enzymes.

Unlike conventional approaches that modify naturally occurring proteins, this technology enables de novo protein design from scratch. Dr. Rhys Grinter noted: "Deep learning methods can efficiently design proteins with specific functions, dramatically reducing development costs." Daniel Fox, the PhD student who led the experimental work, emphasized: "It is crucial that these tools are accessible to researchers worldwide — this will drive the democratization of protein design."

Professor John Carroll, Director of the Monash Biomedicine Discovery Institute, praised the project for demonstrating Australia's innovative strength in AI-driven protein design. By combining expertise in structural biology and computational science, the team has built a flexible and highly efficient design system, providing a powerful platform for developing next-generation antibacterial therapies.