Credit: Jackie Niam/Getty Images
Artificial intelligence is already capable of sifting through huge amounts of data to identify promising new antibiotics in existing compounds. Now, researchers have shown that generative AI models can invent antimicrobial peptides (AMPs) from scratch.
“We’re leveraging the same AI algorithms that generate images, but augmenting them to design potent new molecules,” says Professor Pranam Chatterjee of Pennsylvania University in the US.
Chatterjee and collaborators designed the model, AMP-Diffusion, and used it to create 50,000 new antibiotic candidates.
The model is presented in a paper in the journal Cell Biomaterials.
“That’s far more candidate drugs than we could ever test,” says co-author César de la Fuente, also a professor at Penn. “So, we used AI to filter the results.”
This identified 46 promising AMPs which were then synthesised, tested on human cells and used to treat skin infections in mice.
The researchers found that 76% of the tested peptides killed bacteria, including multidrug resistant strains, with low toxicity.
But 2 AMPs were the clear standouts.
They proved to be just as effective as levofloxacin and polymyxin B – which are used to treat antibiotic-resistant bacteria clinically – with no adverse effects.
“It’s exciting to see that our AI-generated molecules actually worked,” says de la Fuente. “This shows that generative AI can help combat antibiotic resistance.”
Antimicrobial resistance (AMR) is one of the greatest challenges facing the health of people around the globe. It was directly responsible for 1.27 million global deaths in 2019 and is projected to result in 39 million deaths between now and 2050.
“We’ve shown the model works, and now if we can steer it to enhance beneficial drug-like properties, we can make ready-to-go therapeutics,” says Chatterjee.
“Ultimately, our goal is to compress the antibiotic discovery timeline from years to days,” adds de la Fuente.