Details Matter in Structure-based Drug Design
A detailed understanding of molecular recognition in protein-ligand complexes is essential to translate biostructure information into molecular designs with improved target binding and selectivity. At Roche, we have investigated many aspects of this over the last decade, namely the propensities and geometric preferences of different classical and non-classical intermolecular interactions,[1-4] intramolecular hydrogen bonds,[5] or small molecule conformations.[6] Data mining in structural databases, such as the small molecule Cambridge Structural Database (CSD) and the Protein Data Bank (PDB), were often at the core of our studies allowing us to combine results derived from experimental data with high-level calculations. The content of the CSD and PDB databases were not only useful for knowledge generation, but provided also the basis for powerful new software tools for biostructure analysis and molecular design.[7-8]
Many of our studies were triggered by a lack of understanding in the structure-activity relationship (SAR) interpretation of project data. Using examples from Roche legacy projects (Phosphodiesterase 10, Cathepsin L), I will highlight some of the insightful excursions from specific project questions to a more general understanding of molecular recognition. In addition, I will touch on our most recent approach to annotate and visualize molecular interactions in a novel and highly interactive way.