Related Experiment Videos
Evaluation of docking functions for protein-ligand docking.
1Department of Physiology & Biophysics, Mount Sinai School of Medicine, New York University, One Gustave Levy Plaza, Box 1218, New York, New York 10029, USA.
Journal of Medicinal Chemistry
|November 2, 2001
Summary
This study evaluates docking functions for drug discovery. Statistical potentials performed poorly compared to molecular mechanics when using exact coordinates, but showed similar performance with crystal structure data.
Area of Science:
- Computational chemistry
- Structural biology
- Drug discovery
Background:
- Docking functions are crucial for molecular docking algorithms.
- Both physics-based and statistical potentials exist, but their comparative performance is debated.
Purpose of the Study:
- To propose and apply an evaluation approach for docking functions.
- To compare the performance of molecular mechanics and statistical potentials in protein-ligand docking.
Main Methods:
- Exhaustive enumeration of docking solutions for a rigid docking process.
- Application of molecular mechanics (AMBER) and statistical potentials.
- Linear discriminant analysis using physicochemical descriptors.
Main Results:
- Statistical potentials were less effective than AMBER when using exact coordinates.
- Similar performance was observed for both potential types when using crystal structure coordinates of analogous ligands.
- Steric effects are critical for successful docking in both potential types.
- Molecular mechanics failures were linked to neglecting desolvation and explicit hydrogen bonds.
- Statistical potentials require careful treatment of steric and dispersive forces.
Conclusions:
- The choice of evaluation data (exact vs. crystal structure coordinates) significantly impacts docking function performance.
- Understanding the limitations of each potential type is key for accurate protein-ligand docking.
- This work provides insights for filtering virtual libraries to improve docking accuracy.