AI-Driven Evolution: Creating Phages to Combat E. coli

Stanford researchers have successfully created nearly 300 phages using DNA sequences generated by the Evo 2 AI model. After extensive laboratory testing, they identified 16 phages with potent E. coli-killing abilities.

The research primarily focuses on the bacteriophage ΦX174, pronounced as “FYE-ex-1-7-4”. Brian Hie, an assistant professor of chemical engineering and Dieter Schwarz Foundation Stanford Data Science Faculty Fellow, developed the Evo 2 model, with bioengineering graduate student Samuel King leading the experimental efforts outlined in the study.

Evo 2: Bringing Phage Genomes to Life in the Lab

Evo 2 is capable of generating new DNA sequences from a small segment of a phage genome. The researchers instructed the model to produce a complete ΦX174 genome in a single pass.

Hie explained, “We tasked the model with generating the entire genome without any additional input. This process yielded thousands of potential genomes, from which we selected sequences for synthesis and subsequent laboratory validation.”

ΦX174 served as an ideal test subject due to its compact genome, containing less than 6,000 base pairs compared to the human genome’s 3 billion base pairs. Despite its size, interpreting a 5,400-character DNA sequence gene-by-gene remains a challenging endeavor.

Some of the phages proposed by Evo 2 exhibited superior fitness in lab tests compared to native ΦX174. This project aimed to demonstrate the model’s ability to create complete viable viral genomes rather than suggesting local DNA modifications.

Prioritizing Candidates for DNA Synthesis

King developed a computational framework to streamline the selection of candidate genomes for synthesis. The framework evaluated characteristics derived from ΦX174 and related phages before finalizing options for experimental validation.

Due to practical constraints in DNA synthesis, the researchers utilized Evo 2 to generate genomes, assess them against predefined criteria, chemically synthesize chosen candidates, and evaluate their performance in laboratory settings.

Hie emphasized that the framework significantly reduced synthesis costs by focusing on the most promising candidates identified by the team.

The process highlighted the operational boundaries of Evo 2, as computational evaluation, chemical synthesis, and lab assays were essential to pinpoint viable phages from the array of possibilities generated.

Combatting Resistance with Phage Mixtures

Given bacteria’s ability to develop resistance to single treatments, the researchers selected multiple E. coli-targeting phages. Mixing phages could hinder bacterial resistance by diversifying the treatment approach.

Hie noted, “If bacteria become resistant to one phage, it compromises the entire treatment. However, a mixture of genetically distinct phages makes it harder for bacteria to evade the entire cocktail.”

According to Stanford’s findings, a phage cocktail comprising the 16 chosen phages swiftly overcame resistance in E. coli strains immune to native ΦX174.

Hie suggested that future work could explore phages targeting methicillin-resistant Staphylococcus aureus (MRSA) and Pseudomonas aeruginosa, a common source of hospital-acquired infections resistant to conventional treatments.

Promoting Open-Source Collaboration

Hie has made Evo 2 available as open-source software, enabling researchers to utilize the model for genome design purposes.

The release sparked discussions on safety and security at Stanford. While acknowledging the potential for misuse, Hie argued that existing pathogens pose a greater risk due to their accessibility and ease of production.

Hie highlighted the role of AI-enabled systems in responding to natural pandemics and countering bioterrorism threats, underscoring the tool’s versatility and associated risks.

King emphasized the creative freedom afforded by Evo 2, opening new scientific avenues for exploration. The next phase aims to expand Evo 2’s capabilities to handle longer and more intricate DNA sequences, with ongoing collaborations on bacteriophage designs.

Additionally, the model’s application to small bacterial genomes could support the development of engineered microbes for various purposes, including chemical, medical, and fuel production. Hie’s focus remains on enhancing genetic innovation and outcome controllability.

For more insights on AI and big data, consider attending the upcoming AI & Big Data Expo events in Amsterdam, California, and London, co-located with leading technology conferences.

Powered by TechForge Media, AI News offers insights into emerging enterprise technologies and event updates. Explore upcoming technology events and webinars here.