The Antares project, a collaborative initiative funded by the German Federal Ministry of Research, Technology and Space, aims to leverage artificial intelligence in the search for new antibiotics. This public-private partnership involves key players like the Fraunhofer ITMP, Helmholtz Centre for Infection Research, and Enamine Deutschland, focusing on developing effective treatments for drug-resistant infections.
The project employs AI models trained on publicly accessible datasets to streamline the identification of novel antibiotics. This approach allows for iterative improvements through high-throughput testing of compounds predicted to possess antibacterial properties. Dr. Bernhard Ellinger from Fraunhofer ITMP emphasizes that this iterative validation process distinguishes Antares from other initiatives, which often rely on single training rounds.
In addressing the global challenge of antimicrobial resistance, the partners highlight the urgent need for new antibacterial agents. Dr. Sven Wagner from Enamine notes that the current market for antibiotic development is severely constrained, with minimal returns on investment for drug developers. Antares aims to bridge these gaps by fostering innovation in neglected areas of medical need through collaboration and advanced research techniques.
The project also incorporates Explainable AI (XAI) to elucidate the structure-activity relationships derived from the models, while generative models will propose new antibiotic structures. Professor Mark Brönstrup from HZI underscores the project's ambition to enhance discovery efficiency and success rates by merging AI with high-throughput experimentation, ultimately aiming to characterize the mechanisms of action for the active compounds identified.