AI tool could speed antibiotic development

June 17, 2026

AI tool could speed antibiotic development

Antibiotic resistance is a growing public health concern. Bacterial infections that were once easy to treat are now hard to kill with existing antibiotics and are causing serious illness. Addressing this problem requires developing new antibiotics that work against resistant bacteria.

One way that scientists are looking for new antibiotics is by studying how living things naturally fight bacteria. Some cells make bacteria-fighting chains of amino acids, called peptides. These peptides kill bacteria by tearing or poking holes in the membrane that surrounds a bacterial cell. Scientists have discovered huge numbers of potential peptide antibiotics. However, it's difficult to predict which ones will be most effective. It's also time-consuming to test all the potential ways to improve a promising peptide to make it into a useable medicine.

An NIH-funded team led by Drs. Jacob Gardner and César de la Fuente at the University of Pennsylvania developed a tool that uses artificial intelligence (AI) to design promising peptides to test in the laboratory. The study was published in Nature Machine Intelligence on May 13, 2026.

De la Fuente and colleagues had previously created a system called APEX. This uses AI to identify potential antibiotic peptides in huge biological data sets. For the new study, the researchers added features to APEX to create a new tool called ApexGo. ApexGo suggests changes to existing peptides that are likely to make them better at attacking and killing bacteria. As more information is added, the system can learn and improve its ideas over time.

To test the system, the researchers started with 10 peptides from extinct organisms. These peptides appeared to be effective against different bacteria. They also had a lot of room for improvement in their effectiveness. ApexGo created 10 optimized peptides from each of the original peptides.

The team ran experiments to test the effectiveness of the new peptides against lab-grown bacteria. They found that 86 of the 100 new peptides had the ability to kill at least one kind of bacteria. Sixty-eight percent of the optimized peptides were better at killing bacteria than the original peptide.

The researchers then selected two of the original peptides and two optimized peptides to test in mice infected with an antibiotic-resistant strain of bacteria. They found that the optimized peptides were more effective at fighting infection than the original peptides. The optimized versions were also as effective as existing powerful antibiotics used against antibiotic-resistant infections.

These results suggest that AI methods like ApexGo have potential for designing new antibiotics. Such methods may be faster than and as effective as current techniques at identifying new candidates.

"ApexGO shows that AI can do more than predict which molecules might work: it can help us improve them," says de la Fuente. "At a time when antibiotic resistance is rising worldwide, we need technologies that help us move faster from an idea to a real therapeutic candidate. ApexGO is an important step toward that future."

-by Laura Manella, Ph.D.

Related Links

• Diabetes boosts antibiotic resistance in mice
• Designing a new antibiotic to combat drug resistance
• Immune boost may protect against multiple hospital-acquired infections
• Using viruses to treat antibiotic-resistant bacterial infections
• Searching for new antibiotics in the human body
• Immune boost fights antibiotic-resistant infections in mice
• Stamping out superbugs: A clear and present danger(link is external)
• Antimicrobial (drug) resistance(link is external)
• Antimicrobial resistance (CDC)(link is external)

References

A generative artificial intelligence approach for peptide antibiotic optimization(link is external). Torres MDT, Zeng Y, Wan F, Maus N, Gardner J, de la Fuente-Nunez C. Nat Mach Intell. 2026;8(5):841-856. doi: 10.1038/s42256-026-01237-5. Epub 2026 May 13. PMID: 42206144.

Funding

NIH's National Institute of General Medical Sciences (NIGMS); Defense Threat Reduction Agency; National Science Foundation.