Anthropic Uses Claude AI to Solve 40-Year-Old Vault Structure Mystery

iconMetaEra
Share
AI summary iconSummary
Anthropic’s AI breakthrough in determining the 3D structure of prokaryotic vaults has generated renewed interest in AI and crypto news. Using Claude AI, the team analyzed 150 billion sequences, identifying homologs across 29 bacterial and four archaeal phyla. Laboratory validation confirmed 44 candidates, with seven vault-like particles successfully assembled in E. coli. One phage (L05) achieved 3.2-Å resolution. AI reduced protein purification time from three days to 90 minutes and generated a complete model in 34 hours, though some errors arose due to data processing issues. This advancement adds momentum to the intersection of AI and crypto news.
The Anthropic team used Claude AI to first resolve the 3D structure of the prokaryotic Vault, screening homologs from 15 billion sequences and successfully assembling seven vault-like particles in E. coli. The bacteriophage L05 achieved a resolution of 3.2 Å and exhibited tunable size properties. AI reduced purification time to 90 minutes and model generation to just 34 hours, but errors occurred due to additive interference and skipped processing steps, highlighting both the efficiency and limitations of human-AI collaboration.

Author and source: AIBase

On November 5, Anthropic announced that its biology team, using Claude, has for the first time resolved the 3D structure of the prokaryotic vault. This vesicle-like complex, discovered in 1986, has remained enigmatic for four decades—no one has clearly determined its function in living cells; similar structures number nearly ten thousand within human cells.

image.png

Breaking through with a needle in a haystack: The team sifted through 15 billion predicted sequences to identify homologs, narrowing candidates from 29 bacterial phyla and four archaeal phyla to 44 for laboratory validation. Seven assembled into dome-like particles in E. coli, and cryo-EM confirmed four complete structures. Among them, bacteriophage L05 achieved a resolution of 3.2 Å—featuring 96 identical protein chains, each with seven wall repeats (compared to nine in humans); the particle measures 54.5 × 47.5 nanometers, is slightly spherical, and has an internal volume 25% larger than its human counterpart. The team then inserted additional wall sequences, increasing the variant’s volume by approximately 20%, demonstrating that wall sequence count can precisely tune particle size.

Claude reduced purification from three days to 90 minutes, but also caused some confusion.

Claude takes on multiple roles: protein searching, DNA design, purification optimization, and electron microscopy data processing. What used to take three days for traditional purification is now reduced to just 90 minutes; the first complete structural model, which previously took weeks to generate, is now produced in 34 hours.

AI also showed its limitations: its suggested additive interfered with DNA delivery, causing researchers to wait two weeks for nothing; during electron microscopy processing, it skipped standard steps, producing inaccurate images that were only corrected through intensive oversight. Although the structure has been resolved, the function of the vault remains enigmatic. Data for L05 and its variants have been deposited in the Protein Data Bank, providing a size-tunable scaffold for cargo encapsulation—human-AI collaboration has accelerated progress, yet also reminded us that models are far from capable of operating independently.

Disclaimer: The information on this page may have been obtained from third parties and does not necessarily reflect the views or opinions of KuCoin. This content is provided for general informational purposes only, without any representation or warranty of any kind, nor shall it be construed as financial or investment advice. KuCoin shall not be liable for any errors or omissions, or for any outcomes resulting from the use of this information. Investments in digital assets can be risky. Please carefully evaluate the risks of a product and your risk tolerance based on your own financial circumstances. For more information, please refer to our Terms of Use and Risk Disclosure.