SRKnot: Predicting RNA pseudoknot structure based on a stem replacement method

The prediction algorithm for the secondary structure of RNA is highly accurate; however, its pseudoknot prediction accuracy is poor. Recent studies have proposed different pseudoknot classes to predict the secondary structure of RNA; however, few studies have achieved satisfactory complexity and accuracy. Here, we propose a new method for predicting RNA pseudoknot structure based on stem replacement in the absence of pseudoknot classes. The proposed method uses an improved heuristic algorithm combined with a minimum free energy algorithm based on a known or predicted RNA secondary structure. The dynamic programming algorithm establishes the pseudoknot stem pool and sorting (stem length and free energy probability model). The free energy model was developed as an objective function using a heuristic algorithm and stem replacement to predict RNA pseudo structure according to RNA types, or by calculating the pseudoknot structural freedom to predict the optimal structure. The sensitivity-specific Matthews correlation coefficient revealed that the proposed method is both more accurate and superior to the commonly used pseudoknot prediction algorithm. Thus, this study provides a new method of predicting RNA function.