Identifying Morfs In Disordered Proteins Using Enlarged Conserved Features

Identifying the short binding regions, which are called molecular recognition features (MoRFs), within intrinsically disordered proteins (IDPs) is the key step for understanding the function of IDPs, for protein structure determination and for drug design. Due to the complexity of IDPs, highly accurate prediction of MoRFs from its amino acid sequence still remains extremely challenging. Here, inspired by the signal processing technology, we proposed a new method which is based on the enlarged conserved features of sequence for MoRFs prediction. In our approach, only the revised position-specific scoring matrix (PSSM) generated from the sequence was used as input feature, and the support vector machine (SVM) was adopted to build the prediction model. Finally, the output prediction scores were processed by an average strategy to further improve the accuracy. When compared with other single model-based methods on the same datasets, our results were very competitive in terms of accuracy with respect to the state-of-the-art methods.