Particle Filter Method to Integrate High-Speed Atomic Force Microscopy Measurements with Biomolecular Simulations

Area of Science:

• Biophysics
• Computational Biology
• Microscopy

Background:

• High-speed atomic force microscopy (HS-AFM) offers real-time, single-molecule observation of biomolecular dynamics under near-physiological conditions.
• HS-AFM's spatiotemporal resolution is limited, while molecular dynamics (MD) simulations provide higher resolution but may contain artifacts.

Purpose of the Study:

• To develop a novel particle filter method for integrating HS-AFM data with coarse-grained molecular dynamics (CG-MD) simulations.
• To enhance the spatiotemporal resolution and accuracy of biomolecular dynamics analysis.

Main Methods:

• A sequential Bayesian data assimilation approach using a particle filter was developed.
• A twin experiment was conducted using a nucleosome model, generating a reference HS-AFM movie from CG-MD trajectory.
• Particle filter simulations were performed with varying particle numbers (8-8192) to assess performance.

Main Results:

• The particle filter method successfully captured large-scale nucleosome structural dynamics compatible with HS-AFM movies.
• Increasing the number of particles in the simulation consistently improved the likelihood of the entire HS-AFM movie.
• The method allowed for the inference of true ionic concentrations and time scale mappings by maximizing the likelihood of the complete movie, not individual frames.

Conclusions:

• The developed particle filter method provides a robust framework for integrating HS-AFM data with MD simulations.
• This approach offers a generalizable strategy for improving the analysis of biomolecular structural dynamics.
• The study highlights the potential for combining experimental and computational techniques to overcome individual limitations.