Scientists use AI to create proteins that kill E. coli

Last year there was a surge in proteins developed by AI, and ultimately used for everything from snakes to cancer. It usually takes decades to create a custom-made protein for a particular disease – it can now be done in seconds.

For the first time, Australian scientists have used artificial intelligence (AI) to produce ready-to-use biological proteins. In this case, antibiotic-resistant bacteria such as E. coli can be killed.

This study has been published in Natural Communicationprovides a new way to combat the growing crisis caused by the antibiotic resistance superbug.

By using AI in this way, Australian science is now joining countries such as the US and China to develop AI platforms that can quickly generate thousands of ready-to-use proteins, paving the way for faster, more affordable drug development and diagnosis that can transform biomedical research and patient care.

The paper is co-led by Dr. Rhys Grinter and Professor Gavin Knott of Snow Medical Fellow, leading a new node-equipped AI protein design program at the University of Melbourne Bio21 Institute and the Monash Biomedicine Discovery Institute.

According to Dr. Grinter and A/Prof. Knott, the AI ​​protein design platform used in this work, is the first in Australia to model the work David Baker (who won the Nobel Prize in Chemistry last year) and develop an end-to-end approach that allows for the creation of a wide range of proteins.

“These proteins are currently being developed as drugs, vaccines, nanomaterials and small sensors, and many other applications have yet to be tested,” Associate Professor Knott said.

In this study, the AI ​​protein design platform used AI-driven protein design tools that are freely available to scientists everywhere. “It's important to democratize protein design so that the whole world can take advantage of these tools,” said PhD Daniel Fox. Students who conducted most of the experimental work for their research.

“Using these tools and tools we develop in-house, we can design proteins to bind specific target sites or ligands as inhibitors, agonists or antagonists, or enzymes with improved activity and stability.”

Currently, according to Dr. Grinter, proteins used to treat diseases such as cancer and infectious diseases are naturally derived and reused through rational design or in vitro evolution and selection.

“These new methods in deep learning allow for efficient de novo design of proteins with specific properties and functions, reducing costs and accelerating the development of novel protein binders and engineered enzymes,” he said.

Since David Baker's work, new tools and software have been developed, including Bindcraft and Chai, which are incorporated into the AI ​​protein design platform co-led by Dr. Grinter and A/Prof. knot.

Professor John Carroll, director of the Monash Biomedicine Discovery Institute, said the new AI protein design program is “speeding the speed of this exciting new modality for designing new therapies and research tools.”

“The program, based at Monash University and the University of Melbourne, is run by a team of talented structural biologists and computer scientists who understand the design process from the end-to-end. This in-depth knowledge of protein structure and machine learning makes it a highly agile program that allows you to regularly onboard cutting-edge tools for AI-Protein design.