Anthropic says its Claude AI model has helped identify a previously uncharacterized enzyme system in bacteriophages, viruses that infect bacteria. The enzyme system discovery by Claude involves a system called array-associated reverse transcriptases (ART), which contains repeating DNA sequences that resemble patterns found in CRISPR. However, researchers do not yet know what ART does or whether it could become a useful biotechnology tool.
The finding of a novel enzyme system is the first result announced from Anthropic’s new life sciences research program, which combines AI-based genomic analysis with laboratory experiments performed by human scientists.
How Did The Claude Enzyme Discovery Happen?
Anthropic asked Claude agents to search a large database of DNA sequences for unusual examples of reverse transcriptases, which led to the enzyme system discovery by Claude. A reverse transcriptase is an enzyme that copies RNA into DNA.
According to Anthropic, about 950 Claude agents spent 21 hours analyzing the data using approximately 210 million tokens. The agents examined more than 200,000 reverse transcriptases, identified about 3,500 candidate systems, and narrowed them to 20 candidates for detailed analysis.
One Claude agent then focused on an unusual reverse transcriptase family and detected a tandem array of repeated DNA sequences nearby. It counted the repeats, measured their spacing, compared the arrangement with known systems, and searched scientific literature before submitting the candidate for human review.
Anthropic researchers subsequently tested the candidate in the laboratory and identified it as a previously uncharacterized enzyme system.
What Is the ART Enzyme System?
The ART enzyme system consists of three main components:
- A reverse transcriptase
- A neighboring partner gene
- A long array of evenly spaced DNA repeats
The reverse transcriptase itself was not newly discovered. Previous research had identified the enzyme in a jumbo phage. The Anthropic enzyme discovery instead centers on Claude recognizing the broader combination of the enzyme, the non-coding repeat array, and an additional protein whose function remains unknown.
Initial laboratory experiments related to the enzyme system discovery by Claude found that the ART repeat array produces multiple distinct short RNAs. This feature has attracted attention because CRISPR arrays also generate RNA molecules that help make CRISPR-Cas systems programmable.
Why Is The Claude Enzyme Discovery Compared With CRISPR?
CRISPR is a natural bacterial defense system that researchers have developed into a powerful gene-editing technology. Its programmable properties have helped transform biological research and the development of gene-editing medicines.
ART shares some structural characteristics with CRISPR-like systems, particularly its repeated DNA sequences. Anthropic also says systems with similar combinations of features have been associated with biological operations involving DNA, such as cutting, copying, and inserting genetic material.
However, the enzyme system discovery by Claude should not be described as the discovery of a new CRISPR system. Researchers have not demonstrated that ART performs the same functions as CRISPR, and its biological role remains under investigation.
What Is Anthropic Life Sciences Lab?
The Anthropic life sciences lab is using AI to search genomic data, generate biological hypotheses, and analyze potential candidates. Human researchers then decide which candidates merit laboratory testing.
Anthropic says all physical laboratory work is currently performed by human scientists. Its Bay Area laboratory operates at BSL-1 and BSL-2 levels and does not handle pathogens capable of infecting humans.
This makes the AI-powered biology research approach different from a fully autonomous laboratory. Claude can rapidly examine large datasets and propose hypotheses, while scientists conduct the experiments needed to determine whether those hypotheses hold up.
Will Claude’s Discovery Lead to New Biotechnology Tools?
The potential applications of ART remain uncertain. Researchers are still investigating its primary biological function, the role of its accessory protein, and how its repeat-derived RNAs work.
If future experiments show that ART can perform programmable DNA operations, it could become relevant to biotechnology research. But that possibility has not been established.
The enzyme system discovery by Claude nevertheless demonstrates how AI could assist with AI drug discovery, genomics, and fundamental biology by examining enormous datasets for patterns that may be difficult to identify manually.
For now, the enzyme system discovery by Claude represents an early-stage scientific finding rather than a new medical or gene-editing technology. Further laboratory research and independent validation will be needed to determine what ART does and whether it has practical applications.
