Property-based and functional characterization of chemical space
Complementary to structure‐based drug design methods are approaches that focus only on smallmolecule ligands independent from their (protein) targets. Chemical structures can be characterized by numerous descriptors, including topological, geometric, electronic, pharmacophoric, and physicochemical. We are interested in developing novel approaches for modeling, visualization and global comparison of large sets of structure‐activity data such as Pubchem and other protein family focused or diverse data files. Our goal is to derive global and local relationships and predictive models that can be used for (protein structure‐independent) ‘target fishing’, identification of promiscuous scaffolds (depending on assay platforms), large‐scale (fast) virtual screening, efficient follow‐up on screening results, library design, and (rational) chemical probe and lead optimization. Similarly to the structure‐based approaches described above, a platform based on ligand structure activity‐derived models is applicable to various biological problems.