Unknotting RNA: A method to resolve computational artifacts

Area of Science:

• Computational Biology
• Structural Biology
• Biophysics

Background:

• RNA 3D structure prediction frequently generates entanglements, which are computational artifacts that complicate models.
• These artifacts often lead to the exclusion of otherwise accurate RNA structural predictions from further study.

Purpose of the Study:

• To present a novel protocol for resolving entanglements in RNA 3D models.
• To preserve the overall 3D fold and structural integrity of RNA models during artifact removal.

Main Methods:

• Utilized the SPQR coarse-grained model and short Molecular Dynamics simulations.
• Implemented energy terms for selective modifications to disentangle structures without significant distortions.

Main Results:

• Successfully resolved over 70% of interlaces and approximately 40% of lassos in 195 entangled RNA models.
• Achieved 81% efficiency in untangling conformations classified as artifacts, with minimal impact on original geometry.
• Demonstrated notable improvement in ClashScore for refined RNA models.

Conclusions:

• The developed protocol offers a viable approach for refining RNA models previously deemed unsuitable due to topological artifacts.
• This method enhances the usability of predicted RNA structures by addressing common computational issues.