ABSTRACT
The function of proteins and their ability to interact with other molecules are determined by spatial amino acid arrangements, which in many cases may not be detectable by protein sequence or global fold analysis [1]. Based on the assumption that similarities in local patterns are responsible for related function or ligandbinding behavior, local feature comparison can be employed to rationalize protein similarity. Protein-binding site comparison can, for example, be used to identify the target proile of drugs [2] and give insights into side effects and toxicity [3, 4]. It can also be used to assess the “druggability” of potential binding sites [5-7] or “deorphanize” proteins with known structure but unknown function [8]. Consequently, comparative pocket analyses support hit and
lead identiication for novel target macromolecules following a chemogenomics approach [9] by uncovering relationships to other, better-studied targets. Numerous approaches have been developed for local modeling and comparison of structural features of proteins (cf. Tables 5.15.3). The focus of these methods lies on the comparison of binding sites for small ligands, also referred to as “binding pockets.” These sites are relevant for protein activity and structure-based drug discovery. Interestingly, they are distinguishable from other protein surface regions by geometric [10-12] and pharmacophoric features (PFs) [13, 14]. Not surprisingly, residue and oligomer compositions of ligand-binding sites differ from expected background frequencies [15, 16]. For instance, tryptophan and histidine residues are observed more than twice as often in binding sites than elsewhere on protein surfaces [17]. The concepts of binding site comparison can also be transferred to protein-protein interfaces (PPIs) or protein-membrane interfaces [18, 19], although many properties of small-molecule-binding sites differ signiicantly from PPI [20].
