A Bayesian approach to single-particle electron cryo-tomography in RELION-4.0

Area of Science:

• Structural Biology
• Biophysics
• Microscopy

Background:

• Electron cryo-tomography (cryo-ET) is crucial for determining macromolecular structures.
• Existing subtomogram averaging methods rely on 3D models, limiting resolution.
• Improving resolution in cryo-ET is essential for detailed structural analysis.

Purpose of the Study:

• To develop a new approach for macromolecular structure determination in cryo-ET.
• To optimize the use of 2D experimental images rather than 3D models.
• To achieve resolutions sufficient for de novo atomic modeling.

Main Methods:

• Optimizing a regularised likelihood target based on 2D images.
• Refining tilt-series alignments and beam-induced motion correction.
• Incorporating Bayesian polishing and CTF refinement principles.
• Implementing the approach in the open-source RELION software package.

Main Results:

• Demonstrated successful structure determination from cryo-ET data.
• Achieved resolutions adequate for de novo atomic modeling.
• Validated the approach across three distinct applications.

Conclusions:

• The new method offers improved resolution for cryo-ET structure determination.
• Directly optimizing 2D image data provides significant advantages.
• The RELION implementation facilitates broader adoption by researchers familiar with single-particle analysis.