We have more than 120,000 protein structures at our fingertips, we are able to quickly access this data by all kind of keyword and sequence queries. But when it comes to querying the real content, the 3D structure, our computational methods are rather limited. Over the past four years we worked on algorithms and database strategies to allow a flexible, efficient, and precise geometric access to protein-ligand interfaces. The method should neither be limited to keywords, precomputed interactions, chemical substructures nor to predefined pharmacophoric points. It should be fast enough to really play with the data with no more than a decent laptop.

Our studies resulted in a novel software piece named PELIKAN, an easy-to-use tool for all kinds of geometric searching in protein-ligand interfaces. No matter whether you are interested in ligand fragments with a specific binding motif, arrangements of side-chain functional groups like aromatic cages, or specific interaction geometries like halogen-phenyl contacts. PELIKAN gives a precise answer, organizes and overlays all structures to the query. The whole story just appeared in JCIM. The tool is part of our software suite which is freely available for academic use (see http://software.molecular-design.net). We preprocessed a series of databases containing protein-ligand complexes related to well-known datasets and interesting target families. More information is available on PELIKAN's web page.