RLDOCKScore: A Scoring Function for RNA-Ligand Docking and Small Molecule Virtual Screening.

Area of Science:

• Computational chemistry
• Drug discovery
• Molecular modeling

Background:

• Virtual screening of RNA targets is key for cost-effective drug discovery.
• Existing RNA-ligand scoring functions excel at pose prediction but lack robust virtual screening evaluation.
• Accurate scoring functions are vital for identifying novel RNA-binding drug leads.

Purpose of the Study:

• To develop an enhanced scoring function, RLDOCKScore, for both RNA-ligand pose prediction and virtual screening.
• To improve upon current scoring functions by incorporating detailed physical effects.

Main Methods:

• Developed RLDOCKScore incorporating stacking, solvation, and conformational flexibility.
• Evaluated RLDOCKScore on 122 RNA-ligand complexes for pose prediction.
• Assessed RLDOCKScore's virtual screening performance on HIV-1 TAR and four riboswitches.

Main Results:

• RLDOCKScore demonstrated superior performance in virtual screening compared to other methods.
• It achieved a top-2% enrichment factor of 25.0 and an AUC of 0.86 for the HIV-1 TAR ensemble.
• RLDOCKScore showed competitive pose prediction accuracy, outperforming most tested functions.

Conclusions:

• RLDOCKScore is a valuable tool for computational drug discovery targeting RNA.
• The function effectively balances pose prediction and virtual screening capabilities.
• RLDOCKScore enhances the identification of novel lead compounds for RNA-based therapeutics.