Prediction of Functional Sites in Proteins by Evolutionary Methods

Functional sites are well-defined regions that are relevant for protein function, and that include characteristic groups of amino acids. These regions may be involved in the interaction between proteins and other molecules, such as other proteins, nucleic acids, small ligands and substrates. Interaction sites have been studied in great detail in representative protein families, and their relationship with natural substrates and drugs has been characterized, as well as their mediation in protein complex formation. In many cases they have been studied in relation to their potential for engineering protein activity. Protein binding sites have also been studied at a more general level by characterizing the typical structure of binding sites, and their general residue preferences. However, it is the relationship between the conservation of sequence features and protein active sites and binding sites that constitutes the basis of the development of prediction methods. The conservation of the chemical characteristics of the amino acids in specific groups of sequences, in the context of large protein families, is a particular method used in a growing collection of methods aimed at predicting protein binding sites at a genomic scale. In this review we analyze these methods, discuss their similarities, and describe a number of key unsolved problems.