Improved metrics for comparing structures of macromolecular assemblies determined by 3D electron-microscopy

Area of Science:

• Structural Biology
• Biophysics
• Computational Biology

Background:

• Advancements in 3D electron microscopy (3D-EM) generate large biomolecular volume datasets.
• Analyzing these complex datasets requires improved methods for comparison, classification, and integration.

Purpose of the Study:

• To develop and evaluate novel scoring functions for comparing 3D-EM volumes.
• To enhance the accuracy and precision of biomolecular structure analysis using 3D-EM data.

Main Methods:

• Development and testing of a suite of scoring functions for 3D-EM volume comparison.
• Utilized benchmark datasets from the Electron Microscopy Data Bank for validation.
• Implemented local mutual information and overlap (LMI_OV) scoring, alongside surface comparison filters.

Main Results:

• Scoring function performance varies with map type, resolution, and volume overlap.
• Incorporating overlap information significantly improves local scoring precision and accuracy.
• The LMI_OV score demonstrates superior overall performance across various conditions.
• Local mutual information outperforms cross-correlation for low-resolution or dissimilar maps.

Conclusions:

• The LMI_OV score is recommended for general-purpose comparison of 3D-EM volumes.
• Surface comparison scores are valuable for low-resolution maps with distinct surface features.
• All developed scoring functions are available in the TEMPy software package.