Claude agents searched a database of DNA sequences for interesting reverse transcriptases, gathered more than 200,000 candidates, and flagged one repeating pattern that a human researcher had never described. Anthropic named it ART. What it does is still unknown.
Anthropic reported on September 23 that Claude agents, given a high-level prompt to search a massive database of DNA sequences for interesting new reverse transcriptases, identified a previously uncharacterized enzyme system. The search ran for 21 hours, used roughly 950 agents, and consumed 210 million tokens. Near the end, one agent flagged a repeating pattern of DNA sequences next to the gene for an unusual reverse transcriptase. Anthropic’s scientists followed up in the lab and named the system array-associated reverse transcriptase, or ART.
ART is found mainly in bacteriophages, the viruses that infect bacteria. It has three parts: a reverse transcriptase, a partner gene beside it, and a long array of evenly spaced DNA repeat sequences. The array resembles a CRISPR array, which holds a bank of different RNA sequences and is what makes CRISPR-Cas systems programmable. Early experiments show the ART array is expressed as a set of distinct short RNAs, which the team reads as a sign that something analogous to CRISPR may be happening. ART has not been shown to be programmable, and the announcement does not claim that it is.
The workflow behind the result follows a pattern the team uses for protein family surveys. Claude first reads relevant literature and reproduces established results from public data, checking that its methods work. It then searches for family members or genomic neighbors that fit no described system, writing a short report proposing a function for each candidate. A second round of critical evaluation follows, and most candidates are eliminated there. Anthropic reports the agents gathered over 200,000 reverse transcriptases, picked out 3,500 new candidate systems, and narrowed those to the 20 most compelling, each written up as a human-readable report. The company says an expert scientist would need weeks to months for this kind of analysis.
The agent that found ART was reading raw DNA next to the reverse transcriptase when it saw a tandem repeat array by eye and described the sequence as spectacular. It then counted the repeats, measured their spacing, compared the layout with known systems, and searched the literature for any earlier report of the pattern. Once convinced the system was new, it filed a report for human review. Anthropic’s scientists then tested the candidate in their Bay Area lab, limited to biosafety levels BSL-1 and BSL-2, with human scientists doing all of the lab work.
Genome mining has a long history: restriction enzymes came from bacterial immune systems, Taq polymerase from a Yellowstone hot spring bacterium, and CRISPR itself was first seen as an unusual repeat sequence. Nearly all of these were found by a person noticing an anomaly. The ART result is an example of an agent doing the noticing. Feng Zhang, a CRISPR pioneer at MIT and the Broad Institute, reviewed the pre-print and called the RNA-repeat arrays intriguing and worth further investigation, without verifying a function for the system.
What remains open is the central question: what ART actually does. Anthropic says experiments to determine its function are underway and chose to publish before they finish. The post gives the token count for the search but no dollar cost, and does not say how many of the other 20 candidates were tested or what they showed. The next thing to watch is the result of the ongoing experiments on ART’s function.
