Researchers at Stanford University have used the Evo 2 generative AI model to create synthetic bacteriophages capable of targeting and killing E. coli bacteria.
Key facts
- •The Evo 2 model generated thousands of candidate genomes, which were narrowed down to 16 phages with strong E. coli-killing activity.
- •The study focused on the bacteriophage ΦX174, which has a genome of fewer than 6,000 base pairs.
- •A cocktail of the 16 selected phages successfully overcame E. coli resistance that was immune to the native ΦX174.
- •Brian Hie has released the Evo 2 model as open-source software for use by the research community.
- •Future research will explore using the model to design phages for MRSA and Pseudomonas aeruginosa.
Stanford University researchers have synthesized nearly 300 bacteriophages using DNA sequences generated by the Evo 2 AI model. Laboratory testing identified 16 of these phages that demonstrated effective E. coli-killing activity. The project, led by assistant professor Brian Hie and graduate student Samuel King, focused on the bacteriophage ΦX174 as a test system to determine if AI can successfully create entire viable viral genomes.
By the numbers
Development and Testing Process
The Evo 2 model generates DNA sequences from a small starting snippet of a phage genome. To manage the thousands of candidates produced, Samuel King developed a computational framework that evaluated potential genomes based on traits from ΦX174 and related phages. This process allowed the team to narrow down candidates for chemical synthesis and laboratory testing, which helped reduce overall synthesis costs.
Addressing Bacterial Resistance
The researchers selected multiple phages for the study because bacteria can develop resistance to individual treatments. By using a cocktail of 16 genetically distinct phages, the team found they could rapidly overcome E. coli resistance that had previously rendered the native ΦX174 ineffective. Brian Hie noted that such mixtures make it significantly harder for bacteria to evade treatment.
Future Applications and Safety
Brian Hie has released Evo 2 as open-source software, allowing other researchers to download and use the model for genome design. Future research aims to apply the model to longer, more complex DNA sequences, including potential targets like MRSA and Pseudomonas aeruginosa. While the release has prompted discussions regarding safety and security, Hie stated that the tool could also assist in responding to pandemics and developing defenses against biological threats.
Advertisement
This article was independently rewritten by ManyPress editorial AI from reporting originally published by Artificial Intelligence News.
