0 comments
With antibiotic resistant infections on the rise, this new breakthrough paves the way for scalable AI-driven phage therapy, while simultaneously creating ethical concerns over biosecurity.
Antibiotics is just one of those discoveries that ended up being fundamental to human history. It turned fatal infections into treatable conditions. In doing so, its inclusion became a necessity in modern healthcare protocols.
However, constant usage or the incorrect prescription of antibiotics in recent decades is now creating a new issue: antibiotic resistance. This happens when bacteria change over time and no longer respond to the medicines developed to either kill them or prevent their growth. In the US alone, this crisis causes at least 2.8 million infections annually, with 35,000 people succumbing to it.
One of the most explored treatments against this is phage therapy where specific bacterial strands are targeted by introducing bacteriophage: a type of virus that replicates within bacteria and eliminates it.
While bacteriophages are abundant on Earth, finding one that is specific to a particular bacterial strain takes a massive amount of time and money. As such, coupled with a long list of regulatory hurdles, phage therapy isn’t easily accessible to many patients and clinicians.
However, a newly published study by researchers from Stanford University and the Arc Institute might just change that. Their breakthrough? Using generative AI to develop synthetic bacteriophages from scratch to fight resistant bacteria.
As part of their investigation, the researchers examined a naturally occurring virus known to infect E.Coli, ΦX174 (pronounced Phi-X-174), which also happens to be one of the first DNA based genomes to ever be sequenced.
Using ΦX174 as a framework, the researchers developed Evo 2, a language model like ChatGPT, trained to understand the biological language of DNA. Through this, they prompted the generation of countless potential DNA sequences for hypothetical phages and synthesised 300 genomes to test against E. coli.
Of the 300, 16 were found to be viable, successfully infecting and destroying E. coli cells.
This breakthrough transforms phage therapy from a slow, unpredictable search for naturally occurring phages, to one that is scalable on demand. This is also made possible because the researchers made Evo 2 open source, distributing its source code and framework freely to the public.
As such, experts worldwide can now rapidly design custom viruses on demand, helping the medical community stay ahead of drug-resistant bacterial strains.
