Researchers Developed Minerva to Decode Bacterial Genomes
The new framework uses language models to identify previously unannotated non-coding elements in prokaryotic DNA.
Updated on Sept. 23, 2026 in Life Sciences
Live Poll
Should artificial intelligence be used more extensively to identify unknown elements in microbial genomes?
Scientists have introduced Minerva, a tool designed to map non-coding elements in prokaryotic genomes. The framework uses genome language models to predict base-pairing interactions that have largely remained ignored by protein-centric research.
Why it matters
Conventional genome annotation is often limited by protein-centric and homology-driven methods, which overlook complex non-coding architecture. Minerva aims to bridge this gap by revealing the functional interactions within intergenic regions.
Minerva analyzed 150 bacterial genomes using genome language models to generate two-dimensional maps of local sequence interactions. It also identified structural extensions in the TwoAYGGAY ncRNA family found in Pseudomonas.
The players
Minerva
This is a new computational framework developed to identify non-coding elements and base-pairing interactions in prokaryotic genomes.
Pseudomonas
This is a genus of bacteria studied by researchers to characterize secondary-structure extensions in its ncRNA family.
The details
The framework employs categorical Jacobian fingerprinting and interaction heads to predict ncRNA base-pairing and motifs. Additionally, the tool successfully detected open-reading-frame signatures at the DNA level and uncovered that UG27 systems template complementary DNA hairpin products.
Timeline
September 23, 2026: The research article was published on biorxiv.org.
The Big Picture
Minerva shifts the scientific discipline away from protein-centric models toward a more comprehensive view of the non-coding landscape, mirroring the evolution seen in the human genome annotation project. By automating the identification of ncRNA, it provides a new pathway to explore genomic dark matter.
This tool could accelerate the discovery of new biological systems and medical treatments by clarifying how non-coding DNA functions. In the long term, these findings may provide stronger foundations for synthetic biology and genomic engineering applications.
The takeaway
Minerva demonstrates the potential of applying language model architectures to decode complex biological sequence data. Researchers can use this framework to explore previously unannotated genomic regions in various bacterial species.
Further reading
Discover more research in Life Sciences.
More information
Review the technical findings in the complete scientific research article.
Source note: This article includes information reported by Biorxiv.
Live Poll
Should artificial intelligence be used more extensively to identify unknown elements in microbial genomes?
