Claude found a CRISPR-like puzzle hidden in phage DNA

Claude found a new enzyme system in the DNA of bacteriophages, the viruses that infect bacteria, Anthropic said in a September 23 post on X, though it does not yet know what the system does. The find is the first result from Anthropic’s new molecular biology lab, where Claude works through data and literature to propose hypotheses and candidate systems, and the company’s biologists review them, test the most promising and do all of the lab work.

The enzyme’s gene sits beside a long array of repeating DNA, an arrangement that looks somewhat similar to CRISPR. That set of features turns up in only a handful of known systems, all of which can cut, copy and paste DNA.

Discoveries of such programmable systems have historically “helped revolutionize medicine,” in Anthropic’s words, and CRISPR is now the foundation of genetic medicines. Whether the new system can be put to similar use will take “much more work” to learn.

CRISPR’s own repeats started out just as unexplained: in 1987 Yoshizumi Ishino’s team at Osaka University, sequencing a gene in E. coli, came across an unusual repeating sequence near it and wrote that “the biological significance of these sequences is not known.” In June 2012 Science fast-tracked the paper in which Jennifer Doudna and Emmanuelle Charpentier showed that Cas9, a DNA-cutting enzyme, could be programmed with a single piece of RNA to cut DNA at a specific sequence.

Anthropic wants to extend the approach behind this find to a broad range of problems, in genomics and in other fields, and is inviting proposals for research questions.