Comparison of several molecular docking programs: pose prediction and virtual screening accuracy.

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Molecular docking is crucial for predicting ligand binding and virtual screening.
• Accurate docking and screening tools are essential for efficient drug discovery.

Purpose of the Study:

• To evaluate and compare the performance of six molecular docking programs.
• To assess docking pose and virtual screening accuracy using diverse protein targets.

Main Methods:

• Six programs (DOCK, FlexX, GLIDE, ICM, PhDOCK, Surflex) were tested.
• Ligand docking to 68 X-ray complexes and virtual screening using the Directory of Useful Decoys (DUD) dataset were performed.

Main Results:

• ICM, GLIDE, and Surflex demonstrated higher accuracy in predicting ligand poses compared to other programs.
• GLIDE and Surflex showed superior virtual screening performance based on ROC AUC and enrichment values.
• Accuracy varied across different protein families, and software parameter optimization significantly impacted results.

Conclusions:

• ICM, GLIDE, and Surflex are recommended for accurate molecular docking and virtual screening.
• Expert knowledge is critical for optimizing docking software parameters and improving prediction accuracy.
• Understanding protein family-specific trends can enhance the application of docking methods.