As an important type of proteins, intrinsically disordered proteins/regions (IDPs/IDRs) are related to many crucial biological functions. Accurate prediction of IDPs/IDRs is beneficial to the prediction of protein structures and functions. Most of the existing methods ignore the fully ordered proteins without IDRs during training and test processes. As a result, the corresponding predictors prefer to predict the fully ordered proteins as disordered proteins.

In this regard, we propose a new method called RFPR-IDP trained with data containing ordered proteins. The predictor is constructed based on the combination of Convolution Neural Network (CNN) and Bidirectional Long Short-Term Memory (BiLSTM), in which CNN can capture the local information or motifs of proteins and BiLSTM can learn the long-term dependence information in both directions of proteins. Experimental results show that RFPR-IDP can effectively reduce the false positive rates and accurately predict IDPs from ordered proteins for real world applications.