Reconstruction of low-resolution molecular structures from simulated atomic force microscopy images

Area of Science:

• Biophysics
• Computational Biology
• Structural Biology

Background:

• Atomic Force Microscopy (AFM) offers nanometer-resolution imaging of biomolecules, capturing their dynamics.
• High-speed AFM provides sequential images, aiding the study of biomolecular function through motion.
• Three-dimensional (3D) structural information is crucial for a comprehensive understanding of biomolecular mechanisms.

Purpose of the Study:

• To develop a computational approach for retrieving 3D structural information from low-resolution AFM images.
• To create a coarse-grained model (Gaussian mixture model) for representing biomolecular structures.
• To enhance the interpretation of biomolecular dynamics observed in AFM experiments by generating 3D models.

Main Methods:

• Utilizing computational modeling to recover 3D information from AFM data.
• Employing a coarse-grained Gaussian mixture model to represent molecular structures.
• Implementing Monte-Carlo sampling to generate candidate models and optimize the similarity between simulated and experimental AFM images.

Main Results:

• The algorithm successfully generated low-resolution 3D models of proteins from simulated AFM images.
• Conformational transitions of proteins were modeled, aiding in the interpretation of AFM data.
• The method demonstrated robust performance across various molecular orientations in AFM images.

Conclusions:

• The developed algorithm can generate 3D low-resolution protein models from AFM images.
• These models facilitate a more detailed interpretation of conformational transitions observed in biomolecular dynamics.
• This hybrid modeling approach is valuable for retrieving comprehensive 3D structural information from AFM data.