11 - Predicting the Evolution, Structure and Function of Proteins from Sequence Information

Abstract

This chapter discusses the prediction of the evolution, structure, and function of proteins from sequence information. The prediction of function of a multidomain protein is a three-step procedure which include prediction of the functions of individual domains, prediction of the protein function from its modular architecture, and prediction of the involvement of the protein in particular pathways or networks. The distinction between enzymatic domains and regulatory domains is partly blurred by identifications of enzyme homologs that possess substitutions of catalytic residues. Gain-of-function mutations provide information about protein function. A sequence-based prediction that the C-terminal coiled-coil regions of dystrophin and the dystrophin-related protein p-dystrobrevin interact has been corroborated experimentally. Solved structures of the inactive states of Src and other Src-family kinases have revealed the intricate associations between the different domains of Src and explain how this kinase is regulated. The intramolecular interactions that maintain Src in its inactive state explain how several mutations can cause constitutive activation of the kinase. It is suggested that although multiple binding functions of Src domains are essential requirements for the physiological regulation of Src's functions, the organismal function of Src is not.