Claude uncovers a novel enzyme system with CRISPR-like repeats

Anthropic has announced a new research group and laboratory in the life sciences. The remit is fundamental biology done with Claude: mining DNA datasets for uncharacterized protein families, generating hypotheses at scale, and testing them in the lab.

How past discoveries began

The authors note that many breakthroughs started with a scientist noticing something odd in nature's molecular diversity. Restriction enzymes, which cut DNA at specific sites, were found in bacterial immune systems and launched the biotech industry. Taq polymerase, which copies DNA at high temperatures, was isolated from a bacterium in a Yellowstone hot spring and became the basis of PCR. CRISPR itself was first spotted as an unusual repeat sequence in bacterial DNA.

What Claude found

In spring 2026 Anthropic assembled a group to see whether general AI models could systematize and speed up such discoveries. Scientists prompted Claude to search a massive sequence database for interesting reverse transcriptases (RTs) — enzymes that copy RNA into DNA. Human involvement was limited to the initial prompt and the lab work.

Over 21 hours, roughly 950 agents burning through 210 million tokens combed the data, and one of them noticed a repeating DNA pattern beside the gene for an odd-looking RT — a "CRISPR-like repeat array." Much like a human researcher, the agent counted the repeats, measured their spacing, compared them with known systems, and searched the literature before filing a report for review.

The system, dubbed array-associated reverse transcriptases (ART), is found mainly in bacteriophages and has three parts: the RT, an adjacent partner gene, and a long array of evenly spaced DNA repeats. The underlying RT, from a jumbo phage, had been described before, but Claude appears to be the first to notice the system's defining features. Early experiments show the ART array is also expressed as a set of short RNAs.

ART's function is still unknown, and work continues. Feng Zhang, a CRISPR pioneer at MIT and the Broad Institute, called the result an "exciting example" of AI agents contributing to biological discovery.

How the work is done

All lab work is performed by human scientists; the Bay Area lab operates only at biosafety levels BSL-1 and BSL-2 and handles no human pathogens. Claude reads the literature, reproduces known results to validate its methods, hunts for candidates outside described systems, writes short reports, and then critically weeds out most of them. A campaign may end with a single candidate worth testing — or none.

Anthropic has released a preprint with more detail and invites other scientists to bring forward research questions.

Source: Hacker News