De novo drug design
Novel molecules with desired properties from scratch
Computational de novo design enables molecular "scaffold-hopping" from known to new molecules. We develop machine learning models and generative AI to find novel chemotypes with desired properties. Our goal is to provide surprising, innovative ideas for discovering new pharmacologically active molecules. By partnering with the pharmaceutical industry, we apply our models to ongoing drug discovery projects.

Our methodology includes ligand- and receptor-based techniques, alongside AI methods that simultaneously consider multiple properties. This encompasses the synthetic feasibility of computer-generated molecules and their polypharmacology.
We often start with known drugs or pharmaceutically active natural products as templates for deep learning models. Our algorithms then suggest drug-like molecules with motivated scaffold variations. Advanced concepts of structure-activity landscapes and multi-objective design techniques open new avenues for hit and lead finding in drug discovery.
Nature-inspired molecular design and optimization
The search for new molecules by in silico assembly requires smart navigation in chemical space. We develop algorithms that are inspired by problem-solving strategies from nature, for example deep neural leraning, or the path-finding abilities of certain ant species or swarm behavior, and adapt these concepts to chemistry. We also employ game theory and related approaches from the field of artificial intelligence.