Knowledge-based statistical potentials have been shown to be efficient in protein structure evaluation/prediction, and the core difference between various statistical potentials is attributed to the choice of reference states. However, for RNA 3D structure evaluation, a comprehensive examination on reference states is still lacking. In this work, we built six statistical potentials based on six reference states widely used in protein structure evaluation, including averaging, quasi-chemical approximation, atom-shuffled, finite-ideal-gas, spherical-noninteracting, and random-walk-chain reference states, and we examined the six reference states against three RNA test sets including six subsets. Our extensive examinations show that, overall, for identifying native structures and ranking decoy structures, the finite-ideal-gas and random-walkchain reference states are slightly superior to others, while for identifying near-native structures, there is only a slight difference between these reference states. Our further analyses show that the performance of a statistical potential is apparently dependent on the quality of the training set. Furthermore, we found that the performance of a statistical potential is closely related to the origin of test sets, and for the three realistic test subsets, the six statistical potentials have overall unsatisfactory performance. This work presents a comprehensive examination on the existing reference states and statistical potentials for RNA 3D structure evaluation.

RNA pseudoknots are a kind of minimal RNA tertiary structural motifs, and their three-dimensional (3D) structures and stability play essential roles in a variety of biological functions. Therefore, to predict 3D structures and stability of RNA pseudoknots is essential for understanding their functions. In the work, we employed our previously developed coarse-grained model with implicit salt to make extensive predictions and comprehensive analyses on the 3D structures and stability for RNA pseudoknots in monovalent/divalent ion solutions. The comparisons with available experimental data show that our model can successfully predict the 3D structures of RNA pseudoknots from their sequences, and can also make reliable predictions for the stability of RNA pseudoknots with different lengths and sequences over a wide range of monovalent/divalent ion concentrations. Furthermore, we made comprehensive analyses on the unfolding pathway for various RNA pseudoknots in ion solutions. Our analyses for extensive pseudokonts and the wide range of monovalent/divalent ion concentrations verify that the unfolding pathway of RNA pseudoknots is mainly dependent on the relative stability of unfolded intermediate states, and show that the unfolding pathway of RNA pseudoknots can be significantly modulated by their sequences and solution ion conditions.

Double-stranded (ds) RNAs play essential roles in many processes of cell metabolism. The knowledge of three-dimensional (3D) structure, stability, and flexibility of dsRNAs in salt solutions is important for understanding their biological functions. In this work, we further developed our previously proposed coarse-grained model to predict 3D structure, stability, and flexibility for dsRNAs in monovalent and divalent ion solutions through involving an implicit structure-based electrostatic potential. The model can make reliable predictions for 3D structures of extensive dsRNAs with/without bulge/internal loops from their sequences, and the involvement of the structure-based electrostatic potential and corresponding ion condition can improve the predictions for 3D structures of dsRNAs in ion solutions. Furthermore, the model can make good predictions for thermal stability for extensive dsRNAs over the wide range of monovalent/divalent ion concentrations, and our analyses show that the thermally unfolding pathway of dsRNA is generally dependent on its length as well as its sequence. In addition, the model was employed to examine the salt-dependent flexibility of a dsRNA helix, and the calculated salt-dependent persistence lengths are in good accordance with experiments.

Structure evaluation is critical to in silico 3-dimensional structure predictions for biomacromolecules such as proteins and RNAs. For proteins, structure evaluation has been paid attention over three decades along with protein folding problem, and statistical potentials have been shown to be effective and efficient in protein structure prediction and evaluation. In recent two decades, RNA folding problem has attracted much attention and several statistical potentials have been developed for RNA structure evaluation, partially with the aid of the progress in protein structure prediction. In this review, we will firstly give a brief overview on the existing statistical potentials for protein structure evaluation. Afterwards, we will introduce the recently developed statistical potentials for RNA structure evaluation. Finally, we will emphasize the perspective on developing new statistical potentials for RNAs in the near future.

RNA 3D structure prediction remains challenging though after years of efforts. Inspired by the recent breakthrough in protein structure prediction, we developed trRosettaRNA, a novel deep learning-based approach to de novo prediction of RNA 3D structure. Like trRosetta, the trRosettaRNA pipeline comprises two major steps: 1D and 2D geometries prediction by a transformer network; and full-atom 3D structure folding by energy minimization with constraints from the predicted geometries. We benchmarked trRosettaRNA on two independent datasets. The results show that trRosettaRNA outperforms other conventional methods by a large margin. For example, on 25 targets from the RNA-Puzzles experiments, the mean RMSD of the models predicted by trRosettaRNA is 5.5 Å, compared with 10.5 Å from the state-of-the-art human group (i.e., Das). Further comparisons with two recently released deep learning-based methods (i.e., DeepFoldRNA and RoseTTAFoldNA) show that all three methods have similar accuracy. However, trRosettaRNA yields more accurate and physically more realistic side-chain atoms than DeepFoldRNA and RoseTTAFoldNA. Finally, we apply trRosettaRNA to predict the structures for the Rfam families that do not have known structures. Analysis shows that for 263 families, the predicted structure models are estimated to be accurate with RMSD < 4 Å. The trRosettaRNA server and the package are available at: https://yanglab.nankai.edu.cn/trRosettaRNA/.