Abstract:Prediction of RNA secondary structures is an important problem in computational biology and bioinformatics, since RNA secondary structures are fundamental for functional analysis of RNA molecules. However, small RNA secondary structures are scarce and few algorithms have been specifically designed for predicting the secondary structures of small RNAs. Here we propose an algorithm named "PSRna" for predicting small-RNA secondary structures using reverse complementary folding and characteristic hairpin loops of small RNAs. Unlike traditional algorithms that usually generate multi-branch loops and 5[Formula: see text] end self-folding, PSRna first estimated the maximum number of base pairs of RNA secondary structures based on the dynamic programming algorithm and a path matrix is constructed at the same time. Second, the backtracking paths are extracted from the path matrix based on backtracking algorithm, and each backtracking path represents a secondary structure. To improve accuracy, the predicted RNA secondary structures are filtered based on their free energy, where only the secondary structure with the minimum free energy was identified as the candidate secondary structure. Our experiments on real data show that the proposed algorithm is superior to two popular methods, RNAfold and RNAstructure, in terms of sensitivity, specificity and Matthews correlation coefficient (MCC).

PSRna: Prediction of small RNA secondary structures based on reverse complementary folding method