ABC2A: A Straightforward and Fast Method for the Accurate Backmapping of RNA Coarse-Grained Models to All-Atom Structures

RNAs play crucial roles in various essential biological functions, including catalysis and gene regulation. Despite the widespread use of coarse-grained (CG) models/simulations to study RNA 3D structures and dynamics, their direct application is challenging due to the lack of atomic detail. Therefore, the reconstruction of full atomic structures is desirable. In this study, we introduced a straightforward method called ABC2A for reconstructing all-atom structures from RNA CG models. ABC2A utilizes diverse nucleotide fragments from known structures to assemble full atomic structures based on the CG atoms. The diversification of assembly fragments beyond standard A-form ones, commonly used in other programs, combined with a highly simplified structure refinement process, ensures that ABC2A achieves both high accuracy and rapid speed. Tests on a recent large dataset of 361 RNA experimental structures (30-692 nt) indicate that ABC2A can reconstruct full atomic structures from three-bead CG models with a mean RMSD of similar to 0.34 angstrom from experimental structures and an average runtime of similar to 0.5 s (maximum runtime < 2.5 s). Compared to the state-of-the-art Arena, ABC2A achieves a similar to 25% improvement in accuracy and is five times faster in speed.