Zuckerberg's Biohub unveils AI "world model" of proteins
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Sarah Grillo/Axios
Biohub, the Mark Zuckerberg and Priscilla Chan-funded institute, on Wednesday released what it says amounts to "a world model of protein biology."
Why it matters: Biohub says the system can compress years of protein research into hours or days, potentially speeding up disease research and treatment development.
Driving the news: Biohub says the release includes a protein-structure prediction model, a protein language model and ESM Atlas, a map of 6.8 billion proteins and 1.1 billion predicted structures.
What they're saying: "What we've shown is that these models have learned such a high-fidelity world model of biology that you can design protein interfaces computationally, take them into the laboratory and they function as predicted," Biohub head of science Alex Rives said in a statement.
The big picture: A number of efforts see curing disease as one of the most impactful ways to use AI.
- Isomorphic Labs, a spinout from Google, has raised more than $2 billion to fund AI-based drug discovery.
- In April, OpenAI announced a series of models aimed at helping accelerate biological research, while Anthropic has set up its own life sciences effort.
- Meanwhile, Biohub said in April it was investing a further $500 million as part of a Virtual Biology Initiative to help create better AI simulations of the human body.
Yes, but: This work is still steps away from designing a drug that passes clinical trials and any therapeutic use would require more safety testing.
The bottom line: Biohub is betting AI can make biology more programmable — allowing scientists to test ideas computationally before moving into the lab.
