Anthropic says its biology research lab has used Claude to determine the first 3D structure of a vault particle found in prokaryotic cells. Vaults are large, hollow protein barrels present in cells across the tree of life, and four decades after their discovery, nobody knows what they do.
The team solved the structure with cryo-electron microscopy. The particle comes from a bacteriophage and is built from 96 copies of a single protein, called L05, with a combined mass of about 9.5 megadaltons. Anthropic says the whole project, from screening candidates to solving the structure, took under four months.
Leonard Rome and Nancy Kedersha discovered vaults in 1986. In mammals the shell is made of 78 copies of the major vault protein, and the particle is about three times the mass of a ribosome. Proposed roles include moving molecules between the nucleus and cytoplasm, protecting cells from toxic compounds such as cancer drugs, and supporting innate immunity, but none has been confirmed, and mice without vaults survive.
“Will we find out that they are all doing the same thing? Or will we see an evolutionary development of vault function? Maybe they are doing the same thing as human vaults, and that is also very interesting.”— Leonard Rome, Vault co-discoverer and adviser to Anthropic
Key facts
- 01Anthropic says Claude helped determine the first 3D structure of a vault particle found in prokaryotic cells.
- 02The phage vault is built from 96 copies of a protein called L05, with a combined mass of about 9.5 megadaltons.
- 03Claude Science searched about 15 billion predicted proteins and found vault relatives in 29 bacterial and four archaeal phyla.
- 04A purification step that conventionally takes about three days was cut to about 90 minutes.
- 05The project ran under four months; a preprint has been submitted to bioRxiv.
The search started with Claude Science, which combed a database of about 15 billion predicted proteins. It found vault relatives in 29 bacterial phyla and four archaeal phyla, present in under 1% of species with a sequenced genome. The team picked 44 proteins to test in a cell-free system, then moved the best performers into living E. coli, where 15 of 24 produced soluble protein.
Purification is where the biggest time savings came. Claude first argued against using size-exclusion chromatography, then proposed loading ruptured-cell contents directly onto the column. That cut a step that conventionally takes about three days to roughly 90 minutes from cells to microscope grid. Seven of the 15 proteins assembled into vault-like particles, and cryo-EM confirmed the vault shape for four of them, two from phages and two from bacteria.
The work ran in an internal multi-agent harness with access to Anthropic's compute, and the cryo-EM analysis used hundreds of GPUs in parallel with the open-source RELION software. A first full model was ready in 34 hours, and final maps arrived about a week after processing began. Anthropic estimates the structural model would normally take weeks.
The phage vault differs from the mammalian version in measurable ways. It is roughly 54 nm long and 45 nm wide, against 67 by 40 nm for mammalian vaults, and encloses about 25% more volume. Its wall protein has seven repeating segments instead of nine, its end caps are flatter, and the cap closes in three steps, like a tiered dome.
Anthropic also made 15 altered versions of the protein. Some single changes near the cap prevented the particle from closing, producing open, rosette-like particles. Duplicating two wall segments to reach nine, matching the human count, increased the enclosed volume by about 20%. Claude also flagged a small molecule sitting in a pocket on each of the 96 protein copies whose shape resembles coenzyme A, a match Anthropic says is not yet confirmed.
Not every Claude decision helped. A Claude-written protocol included a growth-medium additive that blocked DNA transfer, which cost the team two weeks.
The work is credited to Januka Athukoralage, Peter Yoon, Nicholas Perry, Matthew Durrant and Anthropic's Life Sciences Team. Rome advised on the research as a paid consultant and provided feedback on the manuscript, and a preprint has been submitted to bioRxiv.
The vault project follows Anthropic's October 20 report on how Claude designed protein binders against scorpion neurotoxins.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
