The RNA-Puzzles Assessments of RNA-Only Targets in CASP16

Area of Science:

• Computational Biology
• Structural Biology
• Bioinformatics

Background:

• RNA 3D structure prediction is crucial for understanding RNA function.
• The RNA-Puzzles collaborative effort, established in 2011, focuses on improving prediction accuracy.
• Existing evaluation metrics may not fully capture the nuances of RNA structure.

Purpose of the Study:

• To apply RNA-Puzzles' automatic evaluation protocols to the 2024 CASP16 competition.
• To assess the performance of RNA 3D structure prediction methods.
• To identify limitations in current prediction and evaluation strategies.

Main Methods:

• Utilized RNA-Puzzles' established protocols for automatic evaluation.
• Included stereochemical parameters, Watson-Crick and non-Watson-Crick pairs, and base stacking in scoring.
• Assessed global parameters like RMSD, TM-score, GDT, and lDDT.
• Applied protocols to targets from the 2024 CASP16 competition.

Main Results:

• Evaluation included standard global and detailed local structural parameters.
• Several challenging targets were noted due to size or multimerization.
• Predictions strong in secondary structure did not guarantee chemical precision or correct identification of non-Watson-Crick interactions.

Conclusions:

• Accurate RNA 3D structure prediction requires more than just good overall folds.
• Chemical precision and correct identification of non-Watson-Crick interactions are vital for valid architecture and function.
• Further refinement of prediction and evaluation methods is necessary for complex RNA molecules.