PlanarFold: a coarse-grained molecular dynamics model of RNA in two-dimensional space

RNAs serve versatile functional roles by virtue of their structures and dynamics. RNA computational models are typically tailored to either perform structural modeling or solve a specific class of folding problems. Here, we present PlanarFold, a coarse-grained RNA model that integrates molecular dynamics simulation in two-dimensional space with dynamic programming to explore the diverse dynamic behaviours of RNAs, achieving a speedup of more than four orders of magnitude compared with all-atom molecular dynamics models. We demonstrate that, at the secondary structure level, PlanarFold quantitatively reproduces experimental results across diverse scenarios, including the native secondary structures, thermodynamics and kinetics, mechanical properties, and co-transcriptional and de novo folding pathways. The conformational dynamics revealed by PlanarFold can provide mechanistic insight into how RNAs perform or lose functions, and offer potential targets for mutagenesis and therapeutics design, as well as guide the development of RNA-based devices.