The First Bacteriophage Generated by AI

Figure 1. The diagram above depicts the genomic and structural architecture of ΦX174-like bacteriophages, which specifically target E.coli.



​   Genomes encode complex interactions that express vital biological functions. With the assistance of two Artificial Intelligence (AI) models, viable bacteriophage genomes have been generated by computational biologist Brian Hie and other colleagues from Stanford University. A mixture of these AI-generated bacteriophages prevented virus-resistant Escherichia coli (E.coli) from growing. Hie and his colleagues leveraged two AI genome language models, Evo 1 and Evo 2, to develop entire genome sequences with realistic genetic structure and desirable host characteristics, using the lytic phage ΦX174 as the design template. Bacteriophage ΦX174 was the first DNA-based genome ever. It was sequenced back in 1977 and has been well researched, with a wide range of information published on it.

   Experimental testing of the AI-generated genomes by Evo 1 and Evo 2 produced approximately 300 potential phage genomes, but only 16 of them were viable with substantial evolutionary originality. These 16 phage genomes produced viable viruses that could infect E.coli. Multiple phages demonstrated higher fitness than ΦX174 in growth and lytic strength. Furthermore, a combination of the AI-generated phages rapidly overcame ΦX174-resistance in three E.coli strains, highlighting the potential for designing phage therapies against rapidly evolving bacterial pathogens.



Links:

https://www.sciencenews.org/article/ai-genome-bacteria-phage

https://arcinstitute.org/news/hie-king-first-synthetic-phage



Tags: #AI-generated #Bacteriophage #Antibiotic-resistant #Bacterial-infection #E.coli