Clustering molecular dynamics trajectories for optimizing docking experiments.

Area of Science:

• Computational chemistry and structural biology.
• Drug discovery and computational intelligence.

Background:

• Molecular dynamics (MD) simulations are vital for rational drug discovery but computationally expensive for large-scale virtual screening.
• Clustering algorithms are used to reduce MD trajectory dimensionality, but novel approaches are needed.

Purpose of the Study:

• To develop a novel methodology for clustering entire MD trajectories using substrate-binding cavity features.
• To optimize virtual screening and docking experiments in a cloud-based environment.

Main Methods:

• Clustering entire MD trajectories based on structural features of the receptor's substrate-binding cavity.
• Utilizing the k-means algorithm with cavity features as input.
• Validating the methodology using interactions between 20 ligands and a 20 ns MD trajectory of a fully flexible receptor (FFR) model.

Main Results:

• A novel methodology for clustering MD trajectories was developed and validated.
• The clustering was optimized using three validity criteria.
• The approach successfully selected representative structures tailored for specific ligands.

Conclusions:

• Clustering MD trajectories using substrate-binding cavity features is a promising technique for optimizing virtual screening.
• This method enhances the feasibility of rational drug discovery by reducing computational costs.
• The approach accurately selects relevant structural ensembles for ligand-based docking.