Protein structural motifs: identification, annotation and use in function prediction

While functional motifs are commonly detected and studied in protein sequences, few three-dimensional (3D) motifs, that is sets of residues spatially close in three dimensions but not necessarily adjacent in the sequence, have been identified so far, mostly through manual approaches. However, structural motifs may reveal novel and important functional sites and allow the detection of evolutionary relationships that are often unrecognizable when the linear amino acid sequence is inspected. The occurrence of similar 3D motifs in unrelated protein structures can in principle allow the transfer of functional annotation and highlight interesting examples of independently evolved functional sites (convergent evolution). Furthermore, the increasing number of experimentally solved protein structures arising from structural genomics projects, oftentimes poorly annotated, requires structure-based methods for fast and reliable functional inference. Such methods generally rely on matching regions of the query proteins with structural motifs associated with known biochemical functions. The systematic and, whenever possible, automatic identification and annotation of new 3D motifs is an important challenge in bioinformatics. In particular, the compilation of a large and robustly annotated collection of structural motifs, conceptually similar to the several existing resources for sequence motifs (e.g. PROSITE (Hulo, et al. 2006), ELM (Gould, et al.), MmM (Balla, et al. 2006), would fill a substantial void in the field. This chapter will cover several aspects of structural motifs and discuss both the approaches aimed at detecting and the procedures used to associate them to a function. The main issues and limitations of the methodologies based on 3D motif for function prediction will also be discussed.