Accurate prediction of secondary structure of tRNAs

Abstract

RNA structure plays a key role in understanding its biological function. Currently, the most successful de novo methods for RNA secondary structure prediction are based on free energy minimization. Existing free energy minimization methods focus more on building energy calculation models and optimizing energy parameters while paying less effort on developing structure space search algorithm. This study proposes a de novo method, namely DPARSS, for RNA secondary structure prediction. DPARSS proposes a novel dynamic programming algorithm for structure space search and intends to generate predictions with minimal free energy. An evaluation on a benchmark dataset consisting of tRNAs indicates that the DPARSS method achieves better predictive performance than existing free energy minimization methods. Additionally, case studies suggest that the improvement achieved by the DPARSS algorithm is likely due to the fact that the predictions generated by DPARSS algorithm form more base pairs, contain less mismatches and form longer base pair segments.