Agent-run genome survey identifies array-associated reverse transcriptases in jumbo phages
Claude agents surveying reverse transcriptase genes across 1.9 billion metagenomic protein clusters flagged a family of jumbo-phage reverse transcriptases paired with an array of repeated non-coding DNA units and a dedicated partner gene, an arrangement the authors say fits no described reverse transcriptase class.
- Lab
- Anthropic
- Model
- Claude Mythos 5
- Field
- Biology
- Date
- 2026-09-23
- Human collaborators
- Peter H. Yoon, Januka S. Athukoralage, Emmanuel Ameisen, Eric Kauderer-Abrams, Nicholas T. Perry, Matthew G. Durrant
What was found
A harness of Claude Code agents, each task split between a worker and a reviewing supervisor, ran from a human-written brief asking for new partner-gene associations of reverse transcriptases. Over 119 tasks, 949 agent sessions and 21.5 hours without human intervention, the agents built their own profile models, recovered about 200,000 reverse transcriptase clusters and filed 19 reports. The defining observation was off-brief: a worker reading the raw DNA upstream of a phage reverse transcriptase noticed a tandem repeat array, checked it against known elements including CRISPR arrays, and reported it. The authors then characterized the family in interactive sessions, finding 95 related clusters, 28 with arrays of 3 to 21 repeats of 15 to 49 nucleotides separated by 120 to 220 nucleotide spacers, a clade beside retrons, and no nearby cas genes. Published infection RNA-seq for Staphylococcus phage SA1 shows the arrays highly expressed, and the authors' own small-RNA sequencing of the SA1 array expressed in E. coli shows it processed into discrete unit RNAs.
Novelty check
The preprint states that the family belongs to no described reverse transcriptase class, and records the nearest prior work itself: the 2021 genome report for the jumbo phage MarsHill identified this reverse transcriptase and proposed an upstream non-coding RNA but described neither the repeats nor the partner gene, and arrays beside the unrelated UG27 reverse transcriptases were reported in a September 2026 bioRxiv preprint, which the authors read as the architecture arising twice. So the enzyme itself was not new; the claim is the array and partner that define the family. The registry was searched on 2026-09-24 for reverse transcriptase, retron, CRISPR and jumbo with no hits. A web search for array-associated reverse transcriptase returned only the announcement and its press coverage, and a search for jumbo-phage reverse transcriptases with tandem repeat arrays returned the DRT and retron literature, including the 2025 finding that phage infection triggers reverse transcription of a non-coding RNA into a repetitive gene. That mechanism builds repeats by reverse transcription rather than carrying them in the genome, which looks distinct, but a specialist should confirm it.
Caveats and known objections
A lab announcement about the lab's own model, so it enters at claimed under editorial rule 1. The preprint is hosted on Anthropic's own site, not a preprint server, and is unreviewed. It carries no data or code availability statement, and the session transcripts it quotes are not published, so the account of which step the agents took rests on the lab's word. The authors state the biological limit plainly: they have not shown that the reverse transcriptase is active, that the unit RNAs are its substrates, or what the system does for the phage, so what is established is a sequence-level family with transcription data, not a characterized enzyme system. Autonomy is graded collaborative, the weaker reading under editorial rule 5. An agent made the observation that defines the family, reading DNA outside what the brief asked for, but humans wrote the brief, chose which of the 19 reports to pursue, defined the family in interactive sessions with Claude and ran the experiments, and the only record of the agents' part is transcripts the lab has not published. The announcement names no model; the model field comes from the preprint. Feng Zhang, quoted in the announcement, is thanked in the preprint for reviewing an early copy, so his remark is not independent commentary. A top Hacker News comment argues for a plainer framing, that Claude identified an undescribed genomic arrangement around a known retron-like reverse transcriptase, which matches the novelty check. The preprint also attributes the recognition to specific internal Mythos 5 signals that respond to repeated DNA; that is a claim about the model, not the biology, and is not graded here.
Nobody outside the lab has checked this yet.
Reading the primary source closely enough to say whether it supports the claim counts as a check, and you are credited on the entry.
Or on GitHub: submit a check challenge the grade send a correction or send a pull request
Flag this for triage
Signals order the review queue and nothing else. They are never published, and they never move a grade: that takes a citation.
Entry history (1 event)
- AddedEntered the registry graded Claimed and Collaborative.
Entries are never deleted. A grade that does not hold up is downgraded on the record, with the reason beside it.
Community discussion
Graded claimed for verification and collaborative for autonomy. What these mean.
Cite this entry
whataifound.org. (2026). Agent-run genome survey identifies array-associated reverse transcriptases in jumbo phages. whataifound.org: A Registry of AI Scientific and Mathematical Discoveries. https://whataifound.org/finding/2026-09-23-array-associated-reverse-transcriptasesBibTeX
@misc{whataifound-anthropic-2026-transcriptases,
title = {Agent-run genome survey identifies array-associated reverse transcriptases in jumbo phages},
author = {{whataifound.org}},
year = {2026},
howpublished = {whataifound.org: A Registry of AI Scientific and Mathematical Discoveries},
note = {Result by Anthropic. Verification: Claimed. Autonomy: Collaborative.},
url = {https://whataifound.org/finding/2026-09-23-array-associated-reverse-transcriptases}
}