A comprehensive evaluation of RNA secondary structures prediction methods

Abstract RNAs have important biological functions and the functions of RNAs are generally coupled to their structures, especially their secondary structures. In this work, we have made a comprehensive evaluation of the performances of existing top RNA secondary structure prediction methods, including five deep-learning (DL) based methods and five minimum free energy (MFE) based methods. First, we made a brief overview of these RNA secondary structure prediction methods. Afterwards, we built two rigorous test datasets consisting of RNAs with non-redundant sequences and comprehensively examined the performances of the RNA secondary structure prediction methods through classifying the RNAs into different length ranges and different types. Our examination shows that the DL-based methods generally perform better than the MFE-based methods for RNAs with long lengths and complex structures, while the MFE-based methods can achieve good performance for small RNAs and some specialized MFE-based methods can achieve good prediction accuracy for pseudoknots. Finally, we provided some insights and perspectives in modeling RNA secondary structures.