Predicting Small Molecule Binding Nucleotides in RNA Structures Using RNA Surface Topography.

Area of Science:

• Computational biology
• Structural bioinformatics
• Drug discovery

Background:

• RNA small molecule interactions are vital for drug discovery and inhibitor design.
• Accurate identification of RNA binding nucleotides is crucial for developing effective therapeutics.
• Existing prediction methods struggle with complex RNA structures containing junctions.

Purpose of the Study:

• To develop a novel deep learning model for predicting RNA small molecule binding nucleotides.
• To address the limitations of existing methods in handling complex RNA structures.
• To enhance the accuracy and efficiency of RNA-targeted drug design.

Main Methods:

• Developed ZHmolReSTasite, a deep learning model incorporating spatial correlation via RNA surface topography.
• Characterized nucleotides using sequence and tertiary structure information for high-level representation learning.
• Evaluated model performance on benchmark datasets of simple and complex RNA structures.

Main Results:

• ZHmolReSTasite achieved high precision (72.9% on TE18, 76.7% on RB9) on simple RNA structures.
• Outperformed existing methods by 11.6% in precision on challenging RNA structures with junctions.
• Demonstrated robustness and superior performance across various RNA structures.

Conclusions:

• ZHmolReSTasite effectively predicts RNA small molecule binding nucleotides, including in complex structures.
• The model's use of spatial correlation and RNA surface topography offers significant advantages.
• This tool can accelerate RNA inhibitor design and provide valuable insights for drug discovery.