OpenKnot RNA Pseudoknots: RFDpoly-Generated 3D Structures

OpenKnot RNA Pseudoknots – RFDpoly 3D Structures contains computationally generated RNA 3D structures resembling pseudoknot topologies from the Eterna OpenKnot competition. The dataset was produced by RFDpoly through base pair template-conditioned structure generation and is authored by Andrew Favor of the University of Washington. It includes structures from competition rounds 7a and 7b, filtered by F1-score similarity to target secondary structures.

Use Cases

• Benchmarking RNA 3D structure prediction models based on provided pseudoknot topologies.
• Training generative models for RNA backbone design using the RFDpoly-generated structures.
• Analyzing the relationship between RNA secondary and tertiary structure based on F1-score similarity filtering.
• Validating inverse-folding sequence design methods using structures conditioned on AlphaFold3 predictions.

Strengths

• Structures are generated for specific targets from the Eterna OpenKnot competition, providing a defined context.
• Includes multiple filtering methods (e.g., by F1-score similarity) for different competition rounds.
• References peer-reviewed methods like RFDpoly and AlphaFold3 for generation and prediction.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

University of Washington (Andrew Favor), via paperswithcode.

Collection Method

Computationally generated by RFDpoly through base pair template-conditioned structure generation.