Identifying structural domains in proteins.

Analysis of protein structures typically beginswith decomposition of the structure intomore basic units called structural domains. The underlying goal is to reduce a complex protein structure to a set of simpler, yet structurallymeaningful units, each ofwhich can be analyzed independently. Structural semi-independence of domains is their hallmark: domains often have compact structure that can fold (and sometimes function) independently. The total number of distinct structural domains is currently hovering around one thousand: they are represented by the unique folds in SCOP classification (Murzin et al., 1995) or unique topologies in CATH classification (Orengo et al., 1997). Interestingly, this is what Chothia predicted at a rather early stage of the Structural Genomics era (Chothia, 1992). A significant fraction of these domains is universal to all life forms, others are kingdomspecificandyet others are confined to subgroups of species (Ponting andRussell, 2002;Yang, and Doolittle, and Bourne, 2005). The enormous variety of protein structures is then achieved through combination of various domains within a single structure. This ‘‘combining’’ of domains can be achieved by combining together single domain polypeptide chains within a noncovalently linked structure or by combining domains (via gene fusions/ recombination) on a single polypeptide chain that folds into the final structure (Bennett, Choe, and Eisenberg, 1994). There are benefits to both strategies: the former can be seen as a more economic andmodular approach inwhichmanydifferent structures canbe put together with relatively few components within the cell. The latter, combining a specific set of modules within a single polypeptide chain, ensures that they are expressed together and localized in the same cells or cellular compartments (Tsoka and Ouzounis, 2000). This chapter is concernedwith this latter case: the decomposition of themultidomain polypeptide chain into structural domains.