Study shows AI can design bacteriophages with potential to overcome bacterial resistance
Scientists used genome language models to design complete bacteriophage genomes, with laboratory tests showing that some could infect E. coli and overcome resistance to a naturally occurring phage
In a groundbreaking study published in Science, scientists from Stanford University and the Arc Institute have demonstrated the use of artificial intelligence (AI) to design complete bacterial phage genomes, potentially offering new solutions for treating drug-resistant bacterial infections. The research, which highlights the first successful demonstration of AI designing entire viral genomes, contrasts with previous efforts that focused on designing individual genes or proteins.
By generating thousands of potential phage genomes, chemically synthesizing nearly 300, and testing them in the lab, the researchers identified sixteen viable phages capable of infecting Escherichia coli, the well-known gut bacterium. Utilizing genome language models named Evo 1 and Evo 2, which learn patterns in DNA sequences similarly to how language models understand human language, the researchers were able to create new combinations of genes and regulatory elements.
These phages differed in genome length and contained novel DNA-packaging proteins. Crucially, the AI-designed phages exhibited resilience against bacterial resistance, as demonstrated when a cocktail of these newly created phages quickly overcame resistance to a commonly used bacteriophage, ΦX174, in laboratory experiments. However, the study's implications for human health remain unclear, as the research was conducted in a controlled lab environment using phages infecting bacteria, not humans.
While the findings suggest that AI could enhance the design of phages for treating drug-resistant infections, extensive validation and safety measures are essential before any medical applications can be considered. Experts in a related Science perspective cautioned that the development of AI capable of composing viral genomes raises significant biosafety and biosecurity concerns.
Although the study produced only sixteen viable phage designs out of nearly 300 synthesized, it marks a significant step forward in the potential use of AI in addressing bacterial resistance.
Written by urgent.news from The Hindu Health's reporting — not their text. Machine-written; read the original for the full account.
