New medicines often come with costs and long timelines. In short, it’s expensive and slow. Scientists often spend years testing millions of molecules just to find one that might work. At McMaster University, however, a new AI model is speeding up the process.

This AI looks through existing lists of chemicals and designs brand-new ones. In a recent test, the model, called  SyntheMol-RL, created a new antibiotic that successfully treated drug-resistant staph infections in mice.

AI Explores Chemicals For Antibiotics

Finding a new drug is a bit like finding a needle in a haystack, except the haystack is the size of a planet. There are billions of possible chemical combinations out there. Humans can only test about a million in a lab, but this AI can explore up to 46 billion.

The model uses about 150,000 “building blocks” and 50 chemical reactions to see how they might fit together.

“In the lab, we can build chemical compounds using a set of smaller chemical fragments, which can be stuck together like molecular Lego blocks,” says Assistant Professor Jon Stokes. “SyntheMol-RL configures those fragments in different ways, faster than humans ever could, to create new, larger chemical compounds that should — based on its knowledge — be antibacterial.”

Finding the Right Chemical For Medicine

Just because a chemical kills bacteria doesn’t mean it’s a good medicine. As Stokes points out, bleach and fire also kill bacteria, but you wouldn’t want them in your bloodstream. A real drug has to be able to dissolve in the body and stay non-toxic to human cells.

Older versions of the AI sometimes designed chemicals that killed bacteria but couldn’t dissolve in water, making them useless for patients. The team fixed this by teaching the AI to prioritize “solubility” from the very start.

“There is a lot of conflict between compounds that are antibacterial and compounds that are water soluble,” said Gary Liu, a graduate student who helped develop the model. “In previous studies, filtering for compounds that were both antibacterial and soluble after our prompt often left us with significantly fewer viable drug candidates, so we built solubility right into the generation process and now the model can efficiently design antibiotic candidates with greater clinical promise.”

The researchers are now looking into how exactly their new drug, called synthecin, stops bacteria. While they focus on antibiotics for now, the team says the same technology could eventually help design treatments for cancer or diabetes.