Annotation and prediction of protein function at the domain level

Resources for the annotation and prediction of protein function at the domain level

Most proteins, specially in eukaryotic organisms, comprise multiple domains, which can be seen as the structural, evolutionary and functional units of proteins. Nevertheless, almost all databases and resources for annotating and predicting protein molecular function assign functions to complete protein chains, without distinguishing the particular domain responsible or associated to a given molecular function. Along the time, we have developed the first automatic annotation of proteins at the structural domain level with GeneOntology “molecular function” (GO-MF) terms. That database of annotations, named SCOP2GO, was used for constructing profiles of domains with the same fold and the same function that can be used for matching regions of sequences against them. This is a way of assigning fold and GO-MF function to the domains of newly sequenced proteins. Additionally, the matching of residues of a query protein against conserved positions of the profiles can give clues on its possible functional sites. This methodology has been recently implemented in a web server, COPRED, where any user can paste a single sequence and inspect those predictions in an interactive interface.