Unit-Based de novo Drug Design Using a Library of Protein-Ligand Interaction Sites

Under the Human Genome Project, a rough draft of the human genome will be sequenced in 2000, and almost all in 2003. The next step is to obtain functions of the genes in the genome. Structural genomics and pharmacogenomics are among the promising approaches to elucidate them. A gene function is regarded as the protein function and thus the atomic configurations in functional sites of the protein. The functional sites usually include specific binding sites of ligands, especially small molecules. Therefore the collection and organization of local structural knowledge, including atomic locations of ligands and interacting sites of proteins and ligands, must be a precious resource for computational drug design. We have constructed a library of protein-ligand interaction sites and a scoring system for compounds [1]. We report here an improvement of the library, which now contains the information on substructures and functional groups (functional units) of ligand. We also report the improved scoring scheme. We believe this scoring system can be applied to the drug design by combining the predefined functional units.