A study on the effect of different thermodynamic models for predicting pseudoknotted RNA secondary structures

Ribonucleic Acid (RNA) plays a vital role in many functions of a cell including the synthesis of proteins. Its structure is crucial in allowing RNA to serve its functions. SARNA-Predict-pk, which uses Simulated Annealing (SA), has shown excellent results in predicting the secondary structure of RNA molecules. Since SARNA-Predict-pk identifies low energy structures, it relies on thermodynamic energy models to do its energy calculations. SARNA-Predict-pk incorporates the use of energy models that have been used in the Hotknots 2.0 algorithm for calculations of minimum energy for potential secondary structures. This paper investigates how incorporating new versions of these energy parameters into SARNA-Predict-pk affects the results of SARNA-Predict-pk, specifically in the sensitivity, selectivity and F-measure. The results presented in this paper show that the new parameters in the Dirks-Pierce (DP) and the Cao-Chen (CC) energy models do not improve the results in terms of sensitivity, selectivity and F-measure of the SARNA-Predict-pk algorithm. This supports the use of the older parameters in these energy models when using SARNA-Predict-pk.