Google DeepMind, Google’s flagship AI project, has unveiled an enhanced version of its AI model that predicts not only the structure of proteins but also that of “all molecules of life.”
The work of the new model, AlphaFold 3, will help researchers in medicine, agriculture, materials science, and drug development to test possible discoveries.
Previous versions of AlphaFold only predicted protein structures. AlphaFold 3 goes further and can model DNA, RNA, and smaller molecules called ligands, expanding the model’s capability for scientific use.
The new model improves accuracy by 50%
DeepMind states that the new model shows a 50 percent improvement in prediction accuracy compared to its previous models. “AlphaFold 2 marked a milestone in structural biology and has led to all sorts of amazing research,” said Demis Hassabis, CEO of DeepMind, at a press conference. “AlphaFold 3 is a further step along the path in terms of using AI to understand and model biology.”
AlphaFold 3 has a library of molecular structures. Researchers input a list of molecules they want to combine, and AlphaFold 3 uses a diffusion method to generate a three-dimensional model of the new structure. Diffusion is the same type of system used by image generators like Stable Diffusion to assemble photos.
DeepMind states that Isomorphic Labs, a drug discovery company founded by Hassabis, has been using AlphaFold 3 for internal projects. So far, the model has helped Isomorphic Labs improve their understanding of new disease targets.
Along with the model, DeepMind also provides the AlphaFold Server research platform for free to some researchers. The server, powered by AlphaFold 3, allows scientists to generate predictions of biomolecular structures regardless of their access to computing power.
Hassabis states that the server is available for academic and non-commercial use, but Isomorphic Labs is working with pharmaceutical partners to use AlphaFold models in drug discovery programs.
The company claims that it worked with experts in the field and specialists in biosafety, research, and industry to assess the risks surrounding AlphaFold 3, even before its launch.
