Proteome Functional Site Prediction Based On Sequence Was Established By Constructing A Protein Domain Functional Template Library

Prediction of functional sites from protein sequence is a key step in protein function annotation that might provide important evidences for biology experiment. In this study we built a domain profile database of13590protein sequence modules by regrouping the domain structures at the domain level of the Structural Classification of Protein(SCOP) domain database and generating sequence modules by using the profile hidden Markov model (HMMER). Functional sites of135621member sequences in SCOP were derived using dynamics perturbation analysis (DPA) method as well as the SITE comment line from the Protein Data Bank (PDB) file. Function annotated protein domain profile database (/DPD) was built by mapping the functional sites of each member sequence into their corresponding module profile. Several databases were used to check the efficient of this method and the high quality results proved our method has high feasibility. Our analysis showed that fDPD is an open platform which can absorb information from other databases (such as literature search) to become more general or more specific, thus increase the comprehensiveness or accuracy of the prediction. Our method indicated that similar protein structures might have their functional sites located on similar protein surface areas, despite their sequence variations.fDPD is available upon request to the authors.