RNA-Puzzles is a collective endeavor dedicated to the advancement and improvement of RNA 3D structure prediction. With agreement from crystallographers, the RNA structures are predicted by various groups before the publication of the crystal structures. We now report the prediction of 3D structures for six RNA sequences: four nucleolytic ribozymes and two riboswitches. Systematic protocols for comparing models and crystal structures are described and analyzed. In these six puzzles, we discuss (i) the comparison between the automated web servers and human experts; (ii) the prediction of coaxial stacking; (iii) the prediction of structural details and ligand binding; (iv) the development of novel prediction methods; and (v) the potential improvements to be made. We show that correct prediction of coaxial stacking and tertiary contacts is essential for the prediction of RNA architecture, while ligand binding modes can only be predicted with low resolution and simultaneous prediction of RNA structure with accurate ligand binding still remains out of reach. All the predicted models are available for the future development of force field parameters and the improvement of comparison and assessment tools.

The ability to accurately predict the binding site, binding pose, and binding affinity for ligand–RNA binding is important for RNA-targeted drug design. Here, we describe a new computational method, RLDOCK, for predicting the binding site and binding pose for ligand–RNA binding. By developing an energy-based scoring function, we sample exhaustively all of the possible binding sites with flexible ligand conformations for a ligand–RNA pair based on the geometric and energetic scores. The model distinguishes from other approaches in three notable features. First, the model enables exhaustive scanning of all of the possible binding sites, including multiple alternative or coexisting binding sites, for a given ligand–RNA pair. Second, the model is based on a new energy-based scoring function developed here. Third, the model employs a novel multistep screening algorithm to improve computational efficiency. Specifically, first, for each binding site, we use a gird-based energy map to rank the binding sites according to the minimum Lennard-Jones potential energy for the different ligand poses. Second, for a given selected binding site, we predict the ligand pose using a two-step algorithm. In the first step, we quickly identify the probable ligand poses using a coarse-grained simplified energy function. In the second step, for each of the probable ligand poses, we predict the ligand poses using a refined energy function. Tests of the RLDOCK for a set of 230 RNA–ligand-bound structures indicate that RLDOCK can successfully predict ligand poses for 27.8, 58.3, and 69.6% of all of the test cases with the root-mean-square deviation within 1.0, 2.0, and 3.0 Å, respectively, for the top three predicted docking poses. The computational method presented here may enable the development of a new, more comprehensive framework for the prediction of ligand–RNA binding with an ensemble of RNA conformations and the metal-ion effects.

With rapid advances in computer algorithms and hardware, fast and accurate virtual screening has led to a drastic acceleration in selecting potent small molecules as drug candidates. Computational modeling of RNA-small molecule interactions has become an indispensable tool for RNA-targeted drug discovery. The current models for RNA-ligand binding have mainly focused on the docking-and-scoring method. Accurate docking and scoring should tackle four crucial problems: (1) conformational flexibility of ligand, (2) conformational flexibility of RNA, (3) efficient sampling of binding sites and binding poses, and (4) accurate scoring of different binding modes. Moreover, compared with the problem of protein-ligand docking, predicting ligand binding to RNA, a negatively charged polymer, is further complicated by additional effects such as metal ion effects. Thermodynamic models based on physics-based and knowledge-based scoring functions have shown highly encouraging success in predicting ligand binding poses and binding affinities. Recently, kinetic models for ligand binding have further suggested that including dissociation kinetics (residence time) in ligand docking would result in improved performance in estimating in vivo drug efficacy. More recently, the rise of deep-learning approaches has led to new tools for predicting RNA-small molecule binding. In this review, we present an overview of the recently developed computational methods for RNA-ligand docking and their advantages and disadvantages.

Magnesium ions (Mg 2+ ) are vital for RNA structure and cellular functions. Present efforts in RNA structure determination and understanding of RNA functions are hampered by the inability to accurately locate Mg 2+ ions in an RNA. Here we present a machine-learning method, originally developed for computer visual recognition, to predict Mg 2+ binding sites in RNA molecules. By incorporating geometrical and electrostatic features of RNA, we capture the key ingredients of Mg 2+ -RNA interactions, and from deep learning, predict the Mg 2+ density distribution. Five-fold crossvalidation on a dataset of 177 selected Mg 2+ -containing structures and comparisons with different methods validate the approach. This new approach predicts Mg 2+ binding sites with notably higher accuracy and efficiency. More importantly, saliency analysis for eight different Mg 2+ binding motifs indicates that the model can reveal critical coordinating atoms for Mg 2+ ions and ion-RNA inner/outersphere coordination. Furthermore, implementation of the model uncovers new Mg 2+ binding motifs. This new approach may be combined with X-ray crystallography structure determination to pinpoint the metal ion binding sites.

Ion correlation and fluctuation can play a potentially significant role in metal ion–nucleic acid interactions. Previous studies have focused on the effects for multivalent cations. However, the correlation and fluctuation effects can be important also for monovalent cations around the nucleic acid surface. Here, we report a model, gMCTBI, that can explicitly treat discrete distributions of both monovalent and multivalent cations and can account for the correlation and fluctuation effects for the cations in the solution. The gMCTBI model enables investigation of the global ion binding properties as well as the detailed discrete distributions of the bound ions. Accounting for the ion correlation effect for monovalent ions can lead to more accurate predictions, especially in a mixed monovalent and multivalent salt solution, for the number and location of the bound ions. Furthermore, although the monovalent ion-mediated correlation does not show a significant effect on the number of bound ions, the correlation may enhance the accumulation of monovalent ions near the nucleic acid surface and hence affect the ion distribution. The study further reveals novel ion correlation-induced effects in the competition between the different cations around nucleic acids.