Poster Open Access
GeoMine1 on the ProteinsPlus2 web server enables textual, numerical, and geometrical searches in 1,067,518 ligand-based and predicted binding sites in the Protein Data Bank (PDB).2 Given a protein binding site of interest, individual user-defined patterns can be designed for geometric searching in binding sites on the atomic level. This opens new opportunities for typical challenges, e.g., in the field of protein kinase research.4
In this contribution, we show how GeoMine can be used to screen for rare interactions and exploit these, e.g., for the design of selective inhibitors. Additionally, we designed queries based on known protein-ligand interaction patterns of highly active, but unselective compounds to find which kinase binding sites in the Kinase Ligand Interaction Fingerprints (KLIFS)5 database match these interaction patterns. Subsequently, we explore whether additional ligand interaction anchors might lead to more selective inhibitors. In a final example, we perform a screening for reactive cysteine residues in protein kinase structures to identify them and demonstrate its potential for differentiating between cysteine positions that are frequently occurring in protein kinases and cysteine positions that are specific for kinase families. All these applications of the method show that GeoMine can serve as a highly flexible and comprehensive tool to assist in drug design processes, not only for protein kinases but for a multitude of pharmaceutically interesting targets in the PDB.