Abstract—Molecular recognition features (MoRFs) act asmolecular switches in molecular-interaction network of thecell, and assumed to have relationship with the causes ofmany diseases. The importance of identifying MoRFs indisordered proteins is becoming increasingly apparent. Sofar, only a limited number of experimentally validatedMoRFs is known, and there are few specialized tools foridentifying MoRFs. Existing methods used many predictedresults, such as predicted disorder probabilities, solventaccessibility and B-factors as features for prediction, or usedMoRFs database directly for alignment to assist theprediction; however, their design are complex, and theperformance is also affected largely by other predictors. Inthis study, we proposed a novel method, named asMFPSSMPred (Masked and Filtered PSSM basedPrediction), which adopts a masking method to extract highlocal conservative features, and a filtering method to filterout low local conservative scores in position-specific scoringmatrix (PSSMs) for prediction. All features are extractedfrom the sequences only. We compared our method with atraditional PSSM-based method and 9 other existedmethods on a same test dataset. The experimental resultsshowed that, our method achieved the best performancewith AUC of 0.758. This study demonstrated that: 1) theflanking regions of MoRFs affected the plasticity of MoRFs;2) MoRFs were flanked by less conserved residues; and 3)the revised PSSM was predictive features for identifyingMoRFs.  

Index Terms—MoRFs prediction, disordered proteins,PSSM

Cite: Chun Fang, Hayato Yamana, and Tamotsu Noguchi, "Sequence-Based Prediction of Molecular Recognition Features in Disordered Proteins", Journal of Medical and Bioengineering, vol. 2, no. 2, pp.110-114, 2013. doi: 10.12720/jomb.2.2.110-114