Stanford Researchers Synthesize Phages Using AI: A New Frontier in Bacteriophage Therapy

Stanford University has made strides in the ongoing battle against antibiotic-resistant bacteria with an innovative approach that employs artificial intelligence. Researchers synthesized almost 300 phages from DNA sequences generated by the Evo 2 AI model, ultimately narrowing it down to 16 phages that demonstrated exceptional E. coli-killing properties. This pioneering work offers a glimpse into the future of bacteriophage therapy, a potential solution for combating bacterial infections that resist traditional antibiotics.

Meet the Innovators Behind the Study

At the heart of this groundbreaking research is Brian Hie, an assistant professor of chemical engineering and Dieter Schwarz Foundation Stanford Data Science Faculty Fellow. Hie developed the Evo 2 AI model in collaboration with Sam King, a bioengineering graduate student who led the experimental aspects of the project. Their focus was on the bacteriophage ΦX174, a compact organism with a genome of fewer than 6,000 base pairs—minuscule compared to the 3 billion base pairs found in the human genome.

Evo 2: Generating Phage Genomes

Evo 2 operates by generating novel DNA sequences from a small initial snippet of a phage genome. Hie and his team tasked the model with creating an entire ΦX174 genome in a single pass from left to right, without external modifications. This led to the production of thousands of candidate genomes, from which sequences were chosen for chemical synthesis and subsequent laboratory testing.

The team’s choice of ΦX174 was strategic; its simplicity allowed for more focused research efforts. Despite having a relatively straightforward genome structure, Hie noted that even such a compact sequence poses challenges in interpretation at the gene level.

Streamlining the Selection Process

To manage the vast number of candidate genomes generated by Evo 2, King created a computational framework to streamline the selection process. This framework evaluated numerous genomic traits derived from ΦX174 and its relatives, allowing researchers to focus on the most promising sequences for synthesis.

The combination of computational evaluation, chemical synthesis, and lab experimentation meant the researchers could efficiently allocate resources while ensuring that only the most viable candidates underwent rigorous testing. As Hie explained, this method significantly reduced synthesis costs by concentrating efforts on the phages deemed most likely to succeed.

The Power of Phage Cocktails

One of the most exciting findings from this research is the development of a 16-phage cocktail specifically engineered to target E. coli. The rationale behind using multiple phages is that it mitigates the risk of bacteria developing resistance to a single treatment. Hie emphasized that if bacteria become resistant to one phage, it could nullify the effectiveness of the treatment. However, by employing a diverse mix of genetically distinct phages, researchers can considerably complicate the bacteria’s ability to evade all components of the treatment.

Stanford’s experiments revealed that the phage cocktail successfully overcame bacterial resistance to the native ΦX174 phage, opening new avenues for effective therapeutic interventions.

The Open-Source Initiative

In a significant move to expand the scope of this research, Hie has made Evo 2 available as open-source software. This decision allows other researchers to download and use the model for their own genome design projects. However, this development has sparked important discussions regarding safety and biosecurity, with concerns that the tool could potentially be misused.

Hie, while acknowledging these concerns, argues that existing pathogens pose a more immediate threat due to easier access and production options. He believes that AI-enabled systems have the potential to bolster responses to natural pandemics and serve as defenses against biological threats of human origin.

Future Directions for Evo 2

The research team envisions taking Evo 2 a step further, aiming to extend its capabilities to longer and more complex DNA sequences. Hie and his collaborators at Stanford are also exploring additional bacteriophage designs.

Moreover, there’s potential for targeting even smaller bacterial genomes, which could facilitate the engineering of microbes customized for producing vital chemicals, medicines, or fuels. As Hie noted, the project raises two critical questions: How can researchers achieve greater genetic novelty, and how can they enhance the controllability of their outcomes?

In summary, the Stanford team’s research represents a significant milestone in the application of AI in genetic engineering and bacteriophage therapy. Their innovative extraction of phages using Evo 2 could reshape how we view and treat bacterial infections, paving the way for breakthroughs in medicine and public health.