AI models design novel viruses for the first time

In a groundbreaking development, US researchers have successfully synthesized entirely new viruses, not found in nature, based on designs generated by artificial intelligence. The study, published in the journal Science, details how scientists from Stanford University and the Arc Institute created these novel viruses using DNA sequences produced by AI models named Evo 1 and Evo 2.

These "genome language models" function similarly to large language models but are trained on the genetic code of various organisms, including viruses, bacteria, plants, and animals. The researchers tasked the models with generating complete genomes for viable bacteriophages – viruses that infect and destroy bacteria. Using the well-known bacteriophage ΦX174 as a reference, the AI proposed approximately 700,000 potential designs, from which researchers selected 285 promising candidates.

The synthesized DNA molecules based on these designs were then introduced into E. coli bacteria. The researchers observed that 16 of the bacterial cultures began to show clear signs of viral attack, with the newly created viruses successfully replicating and destroying the bacteria. Notably, some of these AI-designed viruses proved more effective at infecting E. coli than the original ΦX174 bacteriophage.

The study's authors emphasized that the significance of their work lies not in the specific characteristics of the created viruses, but in demonstrating the AI models' ability to design viable, novel biological entities with predetermined traits. This capability, they suggest, could theoretically be applied to design far more complex organisms as the AI models continue to improve. The research opens new avenues in biotechnology while also prompting discussions about biosecurity and the responsible development of AI in biological sciences.

“

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.