Binding Site Similarity Search to Identify Novel Target–Ligand Complexes

This chapter reviews both sequence-based and structure-based computational approaches to link target and ligand spaces. Assuming that similar targets bind similar ligands, shared properties of protein-ligand-binding sites may be exploited to predict the most likely target of a given ligand, predict secondary targets accounting for putative side effects, proile a ligand or a complete compound library against a wide array of targets, and ind novel ligands for previously orphan targets. Potential pitfalls and success stories of several methods will be discussed.