Sequence-based Prediction For The Protein-peptide Binding Residues

Protein-peptide interaction plays an important role in cell regulation and signal transduction,showing close relations with many human diseases,such as tumors and neurodegenerative diseases.The peptides interacted with proteins are often short in length,flexible in structure and weak in binding affinity.These challenge the modem experimental techniques.Thus the computational method has become an alternative way for the investigation of protein-peptide binding events.Given the native protein structure,past decades have witnessed the development of the structure-based methods for the identification of peptide binding events.However,lots of proteins do no known their structures.We aim to develop the sequence-based methods,which only rely on the protein sequences,to identify the peptide binding residues.Inspired by facts that the protein-peptide binding is associated with intrinsic disorder,which was not utilized in the existing peptide-binding residue prediction methods,we introduced intrinsic disorder in feature design for the first time,and proposed the novel ab-initio method SVMpep.Recently,the consensus-based methods,which integrated several methods together,have been widely used in protein folding recognition and the prediction of functionally important sites,we therefore designed an accurate consensus-based method PepBind,by assembling our SVMpep and two existing template-based methods S-SITE and TM-SITE together.The comprehensive assessments show that the ab-initio method SVMpep is obviously more accurate than the existing sequence-based method SPRINT-Seq,due to the introduction of intrinsic disorder.The consensus method PepBind not only outperforms all its components,but also provides much better discriminative quality than the latest structure-based method SPRINT-Str.The success of PepBind is attributed to the hybrid design among three complementary components.To the end,we set up an online webserver for consensus-based method PepBind,which is freely available at: http://yanglab.nankai.-edu.cn/PepBind/.It will facillitat the scientific studies for the researchers interested in protein-peptide interactions.