Automated multi-model reconstruction from single-particle electron microscopy data.

Area of Science:

• Structural Biology
• Biophysics
• Computational Biology

Background:

• Biological macromolecules exhibit conformational and compositional heterogeneity, complicating structural analysis.
• Single-particle electron microscopy (SPEM) faces challenges in resolving multiple structural states from heterogeneous datasets.
• Existing methods struggle to accurately model diverse macromolecular assemblies.

Purpose of the Study:

• To develop a fully automated, unsupervised method for reconstructing multiple three-dimensional (3D) structural models from heterogeneous SPEM data.
• To address the challenge of structural heterogeneity in macromolecular complexes.
• To enable the analysis of molecules in different functional states (e.g., ligand-bound vs. unbound).

Main Methods:

• An unsupervised, multi-stage clustering approach is employed, starting with an initial reference structure.
• The method combines Multivariate Statistical Analysis for intra-Euler angle clustering with a novel approach for inter-Euler angle sorting.
• Iterative multi-model projection matching refines the classification and structural model computation from distinct clusters.

Main Results:

• The method successfully reconstructed multiple distinct structural models from both synthetic and experimental heterogeneous datasets.
• Demonstrated efficacy in analyzing datasets with inherent structural flexibility and different molecular states.
• Validated on diverse biological macromolecules, including those in ligand-bound and unbound forms.

Conclusions:

• The proposed automated method efficiently reconstructs multiple 3D models from heterogeneous SPEM data.
• This approach provides a robust solution for analyzing structural heterogeneity in biological macromolecules.
• It offers a valuable tool for advancing the understanding of dynamic and multi-state biological systems.