Münster (mfm/jg) – Developing new antibiotics that remain effective against resistant pathogens is one of the greatest challenges in modern medicine. Researchers at the University of Münster have now demonstrated how generative artificial intelligence (AI) can support the discovery of new antimicrobial agents. The findings were published in the journal npj Antimicrobials and Resistance of the Nature portfolio.

According to estimates by the World Health Organization (WHO), up to ten million people could die each year from infections caused by antimicrobial-resistant pathogens by 2050 if appropriate measures are not taken in time. “However, the development of new antibiotics is unfortunately extremely time- and resource-intensive,” explains Prof. Dominik Heider, head of the Institute of Medical Informatics at the University of Münster. “Artificial intelligence can provide valuable support in this process.”

Sandra Clemens and Dr. Hannah Franziska Löchel, the two first authors of the study, developed both the COMPASS database containing more than 75,000 known antimicrobial peptides and the AI model AmpGPT2. These peptides can combat pathogens through various mechanisms. AmpGPT2 uses the COMPASS data to generate new candidate molecules with antimicrobial properties.

This approach has already demonstrated promising results: A drug candidate developed with the help of AI showed significant activity against the bacteria Klebsiella pneumoniae and Pseudomonas aeruginosa in laboratory experiments. Both pathogens are classified by the WHO as high-priority antibiotic-resistant threats. “Our results demonstrate that AI can make an important contribution to developing new strategies to combat infections caused by resistant bacteria,” summarizes Clemens. Löchel also emphasizes the significance of the study: “The development of COMPASS provides an important foundation for systematically capturing the vast diversity of antimicrobial peptides and making them accessible for AI-based approaches.”

Link to Nature-publication

Link to the press release of the Faculty of Medicine Münster