FRET-guided selection of RNA 3D structures.

Integrative biomolecular modeling of RNA relies on refined structural collections and accurate experimental data that reflect binding and folding behavior. However, predicting such collections remains challenging due to the rugged energy landscape and extensive conformational heterogeneity of large RNAs. To overcome these limitations, we applied a Förster resonance energy transfer (FRET)-guided strategy to identify RNA conformational states consistent with single-molecule FRET (smFRET) experiments. We predicted 3D structures of a ribosomal RNA tertiary contact comprising a GAAA tetraloop and a kissing loop using three popular RNA 3D modeling tools, namely RNAComposer, FARFAR2, and AlphaFold3, yielding a collection of candidate conformations. These models were structurally validated based on Watson-Crick base-pairing patterns and filtered using an eRMSD threshold. For each retained structure, we computed the accessible contact volume of the Cy3/Cy5 dye pair using FRETraj to predict FRET distributions. These distributions were then compared and weighted against experimental smFRET data to identify conformational states compatible with the observed FRET states. Our results demonstrate that experimental transfer efficiencies can be reproduced using in silico predicted RNA 3D structures. This FRET-guided workflow, combined with structural validation, lays the foundation for capturing the highly diverse conformational states characteristic of flexible RNA motifs.