Enhanced Molecular Dynamics Method to Efficiently Increase the Discrimination Capability of Computational...

Area of Science:

• Computational biology
• Structural biology
• Biophysics

Background:

• Protein-protein docking generates potential binding structures, but scoring functions lack discrimination power.
• Current scoring functions are computationally efficient but limit accuracy in identifying native conformations.

Purpose of the Study:

• To evaluate the effectiveness of molecular dynamics (MD) descriptors for ranking protein-protein docking models.
• To improve the accuracy of identifying the nearest-native bound conformation in protein complex prediction.

Main Methods:

• Utilized short, scaled molecular dynamics (MD) runs to calculate descriptors for putative protein-protein complex structures.
• Applied the HADDOCK tool to generate initial sets of docking conformations for eight protein-protein systems.
• Assessed the ability of MD-derived descriptors to distinguish the native-like conformation from decoys.

Main Results:

• MD-derived descriptors effectively recognized the nearest-native bound conformation among HADDOCK-generated models.
• The proposed method demonstrated improved discrimination capabilities compared to standard heuristic scoring functions.
• Successful application across eight diverse protein-protein systems validates the approach.

Conclusions:

• Short molecular dynamics simulations provide valuable descriptors for enhancing protein-protein docking accuracy.
• This approach offers a more reliable method for ranking docking poses, advancing structural biology and drug discovery.
• The integration of MD descriptors represents a significant improvement over traditional scoring functions in protein complex prediction.