AI Uses Large Genome Models to Engineer New Bacteriophages

A team of researchers has unveiled an artificial‑intelligence platform that can generate genetically distant variants of a bacteriophage known for killing pathogenic bacteria. The system, developed at a leading biotechnology institute, employs deep‑learning algorithms to redesign the viral genome while preserving its lytic function, producing a library of phage candidates that differ substantially from the original strain. In laboratory tests, several of the AI‑designed phages demonstrated robust activity against multiple antibiotic‑resistant bacterial isolates, including strains of *Pseudomonas aeruginosa* and *Klebsiella pneumoniae* that are common culprits in hospital infections.

The creation of genetically diverse phage variants aims to address the rapid emergence of bacterial resistance to single‑type phage therapies. By expanding the genetic landscape of therapeutic viruses, the approach seeks to provide a reservoir of alternatives that can be deployed if resistance arises. The researchers plan to advance the most promising candidates into pre‑clinical safety studies, with the long‑term goal of integrating AI‑engineered bacteriophages into the arsenal of treatments for multidrug‑resistant infections.