Antimicrobial resistance is among the greatest global health threats: an increasing number of bacteria are becoming resistant to existing antibiotics, while the development of new drugs is stagnating due to high costs and regulatory hurdles. To counter this trend, innovative and sustainable therapeutic approaches are urgently needed.

This project focuses on antimicrobial peptides (AMPs) — small proteins of the innate immune system that can combat pathogens through multiple mechanisms. These include disrupting bacterial cell membranes or interfering with essential cellular processes, making AMPs a promising alternative to conventional antibiotics, particularly in the fight against multidrug-resistant bacteria.

A central component of the project is the COMPASS database, the world’s largest collection of antimicrobial peptides. It provides the foundation for the development and further refinement of advanced AI models. Building on this resource, we use AmpGPT, a GPT-based model for generating novel antimicrobial peptides that was derived from the protein language model ProtGPT2. AmpGPT is capable of designing new AMPs from all organisms and with diverse modes of action.

A key objective of the project is the targeted advancement of this AI approach: by fine-tuning AmpGPT with the comprehensive data from COMPASS, we aim to generate antimicrobial peptides that are tailored to predefined modes of action. The focus lies on three core properties: high efficacy against pathogens—especially multidrug-resistant bacteria—low toxicity combined with good solubility, and ease of chemical synthesis. The generated peptides are intended to exhibit proven bactericidal or bacteriostatic effects.

The project is conducted as a collaborative effort with the Institute of Medical Microbiology at the University of Münster, bringing together expertise in bioinformatics, artificial intelligence, and microbiology. In the long term, it aims to enable the targeted AI-driven development of new antimicrobial agents and to open up novel therapeutic options in the fight against resistant infections.

Contact: Sandra Clemens, M.Sc.

Funding reference number: He2/003/26