Automated Force-Field Parametrization Guided by Multisystem Ensemble Averages

Abstract

conformation, dynamics, kinetics, and adaptation. In this study, we use the combination of helices and the dynamic programming algorithm to selectively sample RNA two-dimensional (2D) structures. By calculating the free energies of each sampled structures with the Vfold-derived loop entropic parameters for secondary and pseudoknotted structures, and an empirical energy model for kissing motifs, we predict most stable and metastable structures from sequences. We use the stability of helices to balance the computational efficiency and prediction accuracy. Benchmark test shows that the model improves the prediction of RNA 2D structures with and without crossing base pairs.