Structural information about ribonucleic acid (RNA) is lagging behind that of proteins, in part due to its high charge and conformational variability. Molecular dynamics (MD) has played an important role in describing RNA structure, complementing information from both nuclear magnetic resonance (NMR), or X-ray crystallography. We examine the impact of the choice of the empirical force field for RNA structure refinement using cross-validation against residual dipolar couplings (RDCs) as structural accuracy reporter. Four force fields, representing both the state-of-the art in RNA simulation and the most popular selections in NMR structure determination, are compared for a prototypical A-RNA helix. RNA structural accuracy is also evaluated as a function of both density and nature of input NMR data including RDCs, anisotropic chemical shifts, and distance restraints. Our results show a complex interplay between the experimental restraints and the force fields indicating two best-performing choices: high-fidelity refinement in explicit solvent, and the conformational database-derived potentials. Accuracy of RNA models closely tracks the density of 1-bond C-H RDCs, with other data types having beneficial, but smaller effects. At lower RDC density, or when refining against NOEs only, the two selected force fields are capable of accurately describing RNA helices with little or no experimental RDC data, making them available for the higher order structure assembly or better quantification of the intramolecular dynamics. Unrestrained simulations of simple RNA motifs with state-of-the art MD force fields appear to capture the flexibility inherent in nucleic acids while also maintaining a good agreement with the experimental observables.

中文翻译：

---

最大限度地提高RNA结构的精确度，以对抗残留的偶极偶合。

有关核糖核酸（RNA）的结构信息落后于蛋白质的信息，部分是由于其高电荷和构象变异性。分子动力学（MD）在描述RNA结构，补充来自核磁共振（NMR）或X射线晶体学的信息方面发挥了重要作用。我们检查了经验力场的选择对RNA结构优化的影响，该方法使用针对残留偶极偶合（RDC）的交叉验证作为结构准确性报告基因。比较了四个力场，它们代表了RNA模拟的最新技术和NMR结构确定中最受欢迎的选择，它们是典型的A-RNA螺旋结构。还根据输入NMR数据的密度和性质（包括RDC，各向异性化学位移，和距离限制。我们的结果表明，在实验约束和力场之间存在复杂的相互作用，这表明了两种表现最佳的选择：在显式溶剂中的高保真精制和构象数据库衍生的电势。RNA模型的准确性密切跟踪1键CH RDC的密度，其他数据类型则具有有益但较小的影响。在较低的RDC密度下，或仅对NOE进行精制时，两个选定的力场能够以很少或没有实验RDC数据准确描述RNA螺旋，从而使它们可用于更高级别的结构组装或分子内动力学的更好定量。