en.Wedoany.com Reported - Researchers at Stanford University, using generative artificial intelligence tools, have for the first time designed complete viral genomes capable of self-replication in the laboratory, and subsequently synthesized 16 new viruses that can infect bacteria. This study is regarded as a key advancement in the field of synthetic biology, and its findings have sparked discussions on biosafety concerns.
According to Brian Hie, an assistant professor at Stanford University, generative AI has previously been used to design novel antibiotics, but designing a complete genome from scratch that can replicate within a cell represents a new level of complexity. The AI models used by the researchers, Evo1 and Evo2, operate on principles similar to large language models like ChatGPT—while the latter predict text sequences, the former are trained on the genetic codes of viruses, bacteria, plants, and humans to predict the language of life. The researchers tasked the models with generating a type of virus that infects only specific bacteria, namely phages.
In the laboratory phase, scientists selected 302 of the most promising candidates from the AI designs and synthesized each one, of which 16 phages were confirmed to effectively kill E. coli. Doctoral student Samuel King noted that once the system was confirmed to be functioning properly, the laboratory research progressed rapidly.
Phage therapy is considered a potential approach to combating antibiotic-resistant infections. This study also demonstrates that AI has the potential to design biological structures that do not exist in nature, which falls within the realm of synthetic biology. Hie believes the system has the potential to improve human health by developing new drugs and therapies.
However, related safety concerns have also emerged. Thomas Inglesby and Moritz Hanke of the Johns Hopkins Center for Health Security recently wrote that these findings raise urgent biosafety questions. The researchers stated that the issue is no longer whether generative viral genome design will occur, but whether it can be used without causing serious harm.
To mitigate risks, the Stanford team implemented a series of precautionary measures: viruses capable of infecting complex organisms were excluded from the training database, the research used only phages rather than viruses that infect humans, and all experiments were conducted in secure laboratories. Hie believes that existing safeguards help ensure the technology is used for benevolent purposes.
In terms of scale, the genetic code of a phage consists of approximately 5,400 base pairs, the smallest living cell genome is about 500,000 base pairs, and the human genome is composed of 3 billion base pairs. Hie stated that attempting to construct simpler organisms may require substantial effort, but it is not impossible, and the research team is interested in pursuing this direction.
