AI and synthetic biology for new antibiotics: KERN project funded

A new research consortium at the University of Tübingen is combining artificial intelligence with high-throughput laboratory methods to find antibiotics against resistant pathogens. The project KERN is funded by the Federal Ministry of Research, Technology and Space (BMFTR) under its programme "Application of Artificial Intelligence in Drug Discovery" and runs for three years.

Bacterial genomes contain a vast, largely unexplored reservoir of natural products with antibiotic potential. KERN focuses on lasso peptides, which are small, ribosomally produced peptides whose unusual threaded structure makes them exceptionally stable and allows them to attack bacteria in ways that differ from conventional antibiotics.

The team uses machine learning to search bacterial genomes for promising gene clusters and to design new peptide variants. These are then produced in E. coli, diversified using CRISPR methods, and tested in the thousands against the hospital pathogen MRSA. Results from the laboratory flow back into the computational models, so that predictions improve with each round.

"We envision a dataset of thousands of peptides, where we know for each one how well it is produced in E. coli and how strongly it kills MRSA," says CMFI Board Member Hannes Link. "That is what our group contributes."

The consortium is coordinated by Nico Pfeifer and brings together the groups of Nadine Ziemert (CMFI Board Member), Evi Stegmann, Hannes Link and Chambers Hughes at the University of Tübingen and the University Hospital Tübingen. Promising substances are to be fed into the antibiotic development pipeline of the German Center for Infection Research (DZIF).

“What excites me most about KERN is that we bring together the enormous potential of natural products as antibiotics with the power of artificial intelligence and high-throughput technologies. Genome mining gives us access to a vast reservoir of unexplored molecules, while AI and synthetic biology allow us to explore this diversity as new antibiotic candidates,” says Nadine Ziemert.