Stanford Evo 2 AI model generates phages against E. coli
| Source: AI News (ainews.com)
Tags: Evo 2, Stanford, synthetic biology, bacteriophage, computational biology, biosecurity, phage therapy
Stanford researchers used the Evo 2 AI model to generate and synthesize nearly 300 novel bacteriophage genomes, with 16 showing strong E. coli-killing activity in lab tests — the first time AI created fully functional, previously unknown viruses from learned evolutionary patterns.
Details
A Stanford team led by professor Brian Hie and graduate student Samuel King used Evo 2, a generative AI trained on millions of genomes across all biological domains, to design entirely new bacteriophage genomes from scratch. Using bacteriophage phiX174 as a guide — not a template — the model generated entire phage genomes in a single left-to-right pass, producing novel DNA architectures rather than modifying known sequences. From thousands of AI-generated candidates, a computational screening framework reduced the pool to roughly 300 for chemical synthesis. Of these, 16 showed strong E. coli-killing activity in lab tests. Some of Evo 2's suggested phages showed higher fitness than the natural reference phage phiX174 in laboratory conditions. The practical implications for phage therapy are significant: bacteriophages are increasingly studied as alternatives to antibiotics for treating drug-resistant bacterial infections like MRSA. AI-accelerated genome design could dramatically speed the development of targeted treatments. The biosecurity concern is equally real. Existing regulations govern synthesis of known pathogens; AI that generates functional novel genomes from evolutionary patterns alone sits in a regulatory gray zone that current frameworks do not fully address.