Introduction

---

trRosettaRNA is a deep learning-based algorithm designed for the automated prediction of RNA 3D structures. Utilizing a transformer network, it predicts both 2D inter-nucleotide geometries and the corresponding 3D structure. If the initial 3D structure is not physically plausible, Rosetta energy minimization is employed to refine the model, incorporating restraints from the predicted 2D inter-nucleotide geometries. Blind tests, including CASP15, CASP16 and RNA-Puzzles, demonstrate that the automated predictions made by trRosettaRNA are competitive with those produced by leading human teams. The algorithm has been successfully applied to generate confident structural models for 467 Rfam families that lack known structures. The server is freely accessible for every users, including commericial users. Read more about trRosettaRNA...

The output results include (click here for an example):

• Predicted 3D structure model annotated with both global (molecule-wise) and local (nucleotide-wise) confidence scores (pLDDT)
• Predicted inter-nucleotide contacts/distances
• Predicted secondary structure
• Multiple sequence alignment
• Homologous templates (if exist)
• Similar structures identified by structure alignment (if exist)

Submit

Provide the RNA data (mandatory)

Input a RNA sequence (Click for an example input) or a multiple sequence alignment (MSA) (Click for an example input) below.

Or upload the RNA sequence/MSA file:

Input type: Single sequence MSA Single sequence: Searches for homologous sequences to build an MSA before modeling.
MSA: Skips the MSA search and uses your supplied MSA directly for modeling.

Template release date cutoff:

The template library includes structures released on or before April 1, 2026. Max: 2026-04-01

E-value cutoff for homologous sequence search:

10 1 0.1 0.01 0.001 For single sequence inputs only. A lower E-value finds higher-quality homologs but may reduce accuracy if the MSA becomes too shallow. We recommend the default (10), which performed best overall in our benchmarks.

Use custom secondary structure By default, the secondary structure is predicted by trRNA-SS.

Input your custom secondary structure (in dot-bracket or CT format, click for explanation and an example input).

Other information (optional)

Email: (Optional, where the results will be sent to)

Target name: (Optional, your given name to this target)

Keep my results private (check this box if you want to keep your job private. A key will be assigned for you to access the results)

Submit Reset

News

---

( >> read more ... ).

References

---

Wang et al, The trRosettaRNA server for RNA structure prediction, Nature Protocols, 21: 4705-4728 (2026). (PDF)

Wang et al, Predicting RNA 3D structure and conformers using a pre-trained secondary structure model and structure-aware attention, Nature Machine Intelligence, 8: 722-734 (2026). (PDF)

Wang et al, trRosettaRNA: automated prediction of RNA 3D structure with transformer network, Nature Communications, 14: 7266 (2023). (PDF)

---