AI Designs Novel Viruses Capable of Cell Infection in Lab

Scientists at Stanford University have successfully utilized artificial intelligence to design new viruses capable of killing cells in laboratory settings. This achievement marks the first instance of AI being employed to generate entire viral genomes, the complete set of genetic instructions necessary for an organism's construction.

The research team used AI tools, Evo1 and Evo2, trained on two million bacteriophage genomes. These models, functioning similarly to large language models but trained on genetic codes, proposed thousands of potential genomes. Researchers then synthesized 302 of these AI-generated genomes in the lab and exposed them to Escherichia coli (E. coli) bacteria. Of these, 16 viruses demonstrated the ability to kill the bacteria.

These newly created viruses are bacteriophages, which specifically infect bacteria and do not pose a threat to human, animal, or plant cells. Dr. Brian Hie, a chemical engineer involved in the study, described the process: "In this case, we wanted the model to generate the entire genome end-to-end in a single left-to-right pass. We didn't add anything."

The study, published in the journal Science, was accompanied by commentary from Johns Hopkins experts Dr. Thomas Inglesby and Dr. Maurice Hanke. They highlighted the dual implications of the advance, noting, "Although this is promising for life sciences applications, it also raises urgent biosafety and biosecurity questions. The ability to compose viral genomes using generative AI now exists; the governance to safely steer it does not."

Samuel King, a PhD student in the Stanford lab, expressed excitement about observing the bacteriophages effectively attacking and killing E. coli in petri dishes, noting the appearance of "clear spots" indicating successful infection. The researchers noted in their paper that this work provides a "blueprint for the design of diverse synthetic bacteriophages and useful biological systems at the genome scale."

Scientists involved stated that bacteriophages, possessing some of the smallest known genomes, are particularly accessible for creation. However, they also acknowledged that this research represents a significant step toward AI's application in more complex research endeavors. Dr. Patrick Cai, a researcher at the University of Manchester, commented on the broader implications: "While these are relatively small bacteriophage genomes, the significance extends far beyond phages. It suggests that genome language models are beginning to learn the design principles encoded by evolution, opening the door to AI-assisted genome writing."