It's Time to Create New Antibiotic Classes: Addressing Antimicrobial Resistance Through AI

This is a short series addressing how AI can help in antimicrobial discovery including how the field has moved from a predictive to a generative process, looking at one partnership between a pharmaceutical company and an AI biotech social venture, and how these partnerships may help in reducing antimicrobial resistance.

Phare Bio CEO and Co-founder Akhila Kosaraju, MD, says a major reason we have multidrug resistance today is because we have maintained the existing classes of antibiotics and not created new ones. 

“Many of the drugs, I would argue virtually all of them, have come from the golden era of antibiotic discovery, which was from the 1940s to the 1960s. So those same, dozen or so classes, have also powered extraordinary resistance amongst these bacteria,” she said. “And without new classes, so truly novel mechanisms of action, we're just recapitulating that same resistance, and if anything, just accelerating and amplifying the problem.”

Kosaraju says by utilizing AI they can open up the development process by broadening the chemical compounds to become new potential antibiotics.

“We train our models to not identify identical chemical spaces…So the model is able to start exploring a much vaster potential area of drug discovery than we could ever conduct…It's not just an iterative improvement; it's really creating a possibility that did not exist before.”

Even in this small amount of time AI has been used to create new molecules, that process has changed. Whereas, AI have been using only a small fraction of potential compounds, with generative AI being utilized, Kosaraju says there is much more potential.

“I think now we can exponentiate the amount of coverage of that chemical space to train our models, which then leads to more novel compounds coming out of our platforms,” she said. “With predictive AI, we were really beholden to and limited by the screening libraries that ultimately our compounds were being identified from. And now, with generative AI, we really can take the exploration of our training data set and our models and fully realize the potential of how you can actually drive novelty in the compound structures that are produced.”

To learn more about generative AI or to find out more about AI and antibiotic development, check out the rest of our series.