rsRNASP1: A distance- and dihedral-dependent statistical potential for RNA 3D structure evaluation

Area of Science:

• Computational Biology
• Structural Biology
• Biophysics

Background:

• Knowledge-based statistical potentials are crucial for RNA 3D structure prediction and evaluation.
• Existing potentials show limitations in performance across diverse 3D structure prediction datasets.

Purpose of the Study:

• To develop an improved all-atom statistical potential for RNA 3D structure evaluation.
• To enhance the accuracy of identifying native and near-native RNA structures.

Main Methods:

• Developed rsRNASP1, an all-atom, distance- and torsion-angle-dependent statistical potential.
• Incorporated residue separation and considered torsion angles for backbone, sugar ring, and base to capture local structure.
• Evaluated performance against existing top statistical potentials using various RNA test datasets.

Main Results:

• rsRNASP1 demonstrated superior performance compared to existing potentials in identifying native/near-native RNA structures.
• The new potential showed improved accuracy in ranking decoy structures.
• rsRNASP1 exhibited enhanced performance on the CASP15 dataset, indicating its robustness.

Conclusions:

• rsRNASP1 represents a significant advancement in RNA 3D structure evaluation.
• The inclusion of local structural features via torsion angles improves potential accuracy.
• This potential is a valuable tool for RNA structure prediction and analysis.