HB-CUFIX: Force field for accurate RNA simulations

Accurate modeling of the dynamic structures of ribonucleic acid (RNA) molecules is essential for understanding their biological roles. However, such modeling remains challenging due to limitations in current force fields. This study critically evaluates three RNA force fields, HB-CUFIX, AMBER-χOL3, and AMBER-ROC, comparing their performance against experimental nuclear magnetic resonance and small-angle x-ray scattering data for single-stranded oligonucleotides. Using enhanced sampling techniques, specifically Unified Free Energy Dynamics, we exhaustively sampled the conformational space of tetramer and hexamer RNA sequences, achieving a detailed and thermodynamically converged view of their structural dynamics. Our findings reveal that HB-CUFIX outperforms AMBER-χOL3 and AMBER-ROC, providing near-experimental accuracy in capturing sequence-dependent structural preferences. In particular, HB-CUFIX accurately predicts low energy states for the AAAA and CCCC sequences, favoring A-form helical conformations, while the UUUU sequence adopts an extended, heterogeneous structure. The mixed GACC sequence displays a predominantly A-form helix with flexible terminal residues. These results highlight the significant role of sequence in dictating RNA conformational spaces, which are driven by base stacking interactions and covalent geometry. We also emphasize the importance of enhanced sampling, particularly methods that can handle large numbers of collective variables, in evaluating RNA force fields, as traditional brute-force Molecular dynamics fails to capture the conformational diversity of flexible RNAs. Our study provides a reliable tool for RNA structure prediction and dynamic analysis, supporting future advancements in RNA-targeted research and therapeutic design.