Virtual combinatorial chemistry and in silico screening: Efficient tools for lead structure discovery?*

In this article, an overview of the most common ligand-based in silico screening techniques is given together with an example on the recent successful application of combined use of pharmacophore modeling, database mining, and biological assays. Additionally, a new approach for structure-based high-throughput pharmacophore model generation is presented. The LigandScout program contains an automated method for creating pharmacophore models from experimentally determined structure data, e.g., publicly available from the Brookhaven Protein Databank (PDB). In a first step, known algorithms were implemented and improved to extract small-molecule ligands from the PDB including assignment of hybridization states and bond orders. Second, from the interactions of the interpreted ligands with relevant surrounding amino acids, pharmacophore models reflecting functional interactions like H-bonds or ionic transfer interactions were created. These models can be used for screening molecular databases for similar modes of actions on the one hand, or for screening one single compound for potential side-effects (reversed screening) on the other hand. The implementation was done using the ilib framework, which also formed the basis of the software tool CombiGen, a fragment-based virtual combinatorial library generation program enabling the user to obtain in silico compound collections with high drug-likeness.