Bayesian Nonparametric Analysis of Residence Times for Protein-Lipid Interactions in Molecular Dynamics Simulations

Area of Science:

• Computational biophysics
• Biomolecular simulations
• Statistical mechanics

Background:

• Molecular Dynamics (MD) simulations are crucial for studying protein-environment interactions, especially membrane proteins.
• Quantifying lipid-protein binding kinetics from MD data is challenging due to noise and infrequent long binding events.

Purpose of the Study:

• To develop a robust method for analyzing lipid-protein binding kinetics from MD simulations.
• To characterize residue-resolved residence time distributions and binding processes at different time scales.

Main Methods:

• Application of Bayesian nonparametrics to MD trajectories.
• Unsupervised classification of trajectory frames based on binding process time scales.
• Analysis of cholesterol interactions with six G-protein-coupled receptors (GPCRs) using coarse-grained MD and the MARTINI model.

Main Results:

• Accurate inference of binding kinetics with associated error distributions from MD simulations.
• Identification of distinct binding poses and molecular densities linked to specific kinetic rates.
• Characterization of molecular events underlying a broad range of kinetic rates for cholesterol-GPCR interactions.

Conclusions:

• Bayesian nonparametrics provide a powerful framework for extracting kinetic information from noisy MD data.
• The method enables a deeper understanding of the molecular mechanisms governing lipid-protein binding.
• This approach enhances the quantitative analysis of complex biomolecular interactions in simulations.