Google has spent much of the past year building it Gemini chatbot to counteract ChatGPT, pitching it as a multifunctional AI assistant that can help with work tasks or the digital chores of personal life. More quietly, the company has been working to improve a more specialized one artificial intelligence tool that is already a must-have for some scientists.
AlphaFold, software developed by Google's DeepMind AI unit to predict the 3D structure of proteins, has received a significant upgrade. It can now model other molecules of biological interest, including DNA, and the interactions between antibodies produced by the immune system and the molecules of pathogens. DeepMind added these new capabilities to AlphaFold 3 in part by borrowing techniques from AI image generators.
“This is a big progress for us” Demis HassabisCEO of Google DeepMind, told WIRED ahead of Wednesday's publication of a paper on AlphaFold 3 in the scientific journal Nature. “This is exactly what you need for drug discovery: you need to see how a small molecule is going to bind to a drug, how strongly, and also what else it can bind to.”
AlphaFold 3 can model large molecules such as DNA and RNA, which carry genetic code, as well as much smaller entities, including metal ions. It can predict with high accuracy how these different molecules will interact with each other, Google's research report claims.
The software was developed by Google DeepMind and Isomorphic Labs, a sister company under parent company Alphabet that works on AI for biotechnology and is also led by Hassabis. In January, Isomorphic Labs announced it would collaborate with Eli Lilly and Novartis on drug development.
AlphaFold 3 will be made available via the cloud so that external researchers can access it for free, but DeepMind will not release the software as open source as with previous versions of AlphaFold. John Jumper, leader of the Google DeepMind team working on the software, says it could help better understand how proteins interact and work with DNA in the body. “How do proteins respond to DNA damage; How do they find and fix it?” says Jumper. “We can start answering these questions.”
Understanding protein structures used to require painstaking work using electron microscopes and a technique called X-ray crystallography. Several years ago, academic research groups began testing whether deep learningthe technique underlying many recent developments in AI could predict the shape of proteins simply from the amino acids that compose them, by learning from structures that have been experimentally verified.
In 2018, Google DeepMind revealed that it was working on AI software called AlphaFold to accurately predict the shape of proteins. AlphaFold 2 will be released in 2020 produced results that were sufficiently accurate to create a storm of excitement in molecular biology. A year later, the company released an open source version of AlphaFold that can be used by anyone, along with 350,000 predicted protein structures, including for almost every protein known to exist in the human body. In 2022, the company released more than 2 million protein structures.