Instead of writing essays, this AI is writing genomes that give E. coli a very bad time.
Artificial intelligence has moved beyond taking notes during consultations, reading test results, writing essays and generating art – scientists have now used it to design viruses capable of infecting and killing strains of E. coli.
The Stanford University researchers say their lab-made phages are capable of infecting and killing certain strains ofthe often hard to eradicate E. coli.
Details of the research, which was posted on the preprint servicer bioRxiv this month and is yet to be peer-reviewed, were published in Nature this month.
“This is the first time AI systems are able to write coherent genome-scale sequences,” said Assistant Professor Brian Hie, a computational biologist at Stanford University, California.
The researchers worked with two AI models, called Evo 1 and Evo 2, training them on millions of viral genomes. Out of more than 300 candidate designs, 16 were able to infect E. coli.
In a field where progress often comes in small increments, that number was enough to cause genuine excitement in the lab. The researchers said it showed the potential of AI to design biotechnological tools and therapies for treating bacterial infections.
“Hopefully, a strategy like this can complement existing phage-therapy strategies and someday augment the therapeutics [to] target pathogens of concern,” said Professor Hie.
AI models have already been used to generate DNA sequences, single proteins and multi-component complexes. But designing a whole genome is much more challenging owing to complex interactions between genes and gene replication and regulation processes.
These AI systems were now capable of helping scientists to manipulate highly intricate biological systems, such as whole genomes, Professor Hie told Nature.
Related
“There are many important biological functions that you can only access if you’re able to design complete genomes,” he said.
The researchers said there was still much work to be done.
“Continued progress will likely bring about more generalisable models capable of designing across diverse biological systems with desirable functional properties,” they wrote.
“The rapid pace of improvement in generative biology suggests a future where genome design could become a core biotechnology alongside genome sequencing, synthesis, and editing, possibly enabling the generation of complete living organisms.”
What could possibly go wrong?
Ethical questions linger, as they often do when new technology intersects with biology.
Could an AI that designs helpful viruses also, in theory, design harmful ones?
The researchers stressed that their system was trained only on phages safe for laboratory use. Still, as Professor Kerstin Göpfrich, a biophysicist and synthetic biologist at Heidelberg University in Germany, told Nature, this problem – known as the dual-use dilemma – was not unique to AI, but was always a concern in biology.
“The dual-use dilemma is always there. In biology, if you can build something to help, you can usually imagine a way to misuse it, too,” she said.
“I think this will definitely be a growing field and I’m super excited about it.”
For now, the AI-made viruses represent a step toward new therapies. And while the prospect of “AI-designed life” might sound like the opening scene of a sci-fi movie, the researchers say we’re still a long way from that. Watch this space.
Transmit your AI-optimised virus to Holly@medicalrepublic.com.au.
