Learning Protein-Ligand Unbinding Pathways via Single-Parameter Community Detection.

Area of Science:

• Computational biology
• Molecular dynamics simulations
• Biophysics

Background:

• Understanding protein-ligand interactions is crucial for drug discovery.
• Molecular dynamics (MD) simulations generate complex trajectory data.
• Identifying transition pathways between molecular states from MD data is challenging.

Purpose of the Study:

• To develop a novel, parameter-efficient method for analyzing MD trajectories.
• To identify and characterize pathways of biomolecular processes, such as protein-ligand unbinding.
• To facilitate the identification of reaction coordinates from simulation data.

Main Methods:

• A community detection algorithm is applied to cluster MD trajectories.
• The method requires only a single parameter for trajectory clustering.
• The approach is validated using the streptavidin-biotin complex and the A2a adenosine receptor-inhibitor complex.

Main Results:

• Trajectory clusters generated by the algorithm correspond to distinct pathways.
• The method successfully identifies key transition pathways in complex biomolecular systems.
• The approach aids in pinpointing reaction coordinates for unbinding events.

Conclusions:

• The novel trajectory clustering approach effectively reveals biomolecular pathways.
• This method simplifies pathway analysis in MD simulations.
• It provides a valuable tool for studying protein-ligand (un)binding dynamics.