M-free: scoring the reference bias in sub-tomogram averaging and template matching

Area of Science:

• Structural biology
• Biophysics
• Cellular imaging

Background:

• Cryo-electron tomography (cryo-ET) visualizes cellular proteomes.
• Template matching detects macromolecular complexes in situ.
• Cross-correlation methods suffer from reference bias and lack confidence metrics.

Purpose of the Study:

• To develop a reliable confidence measurement for macromolecular complex detection in cryo-ET.
• To introduce a heuristic to quantify and mitigate reference bias in template matching.

Main Methods:

• A novel M-free score heuristic was developed, analogous to R-free in X-ray crystallography.
• The method measures reference bias in real space using a split mask (9:1 working/testing ratio).
• The heuristic was validated using artificial data, purified macromolecules, and whole Mycoplasma cell tomograms.

Main Results:

• The M-free score accurately quantifies reference bias in template matching.
• The heuristic demonstrated applicability in both template matching and sub-tomogram averaging.
• Successful testing on diverse cryo-ET datasets confirmed the method's robustness.

Conclusions:

• The M-free score provides a reliable measure of reference bias in cryo-ET.
• This heuristic enhances the accuracy and confidence of macromolecular complex identification.
• The method is broadly applicable to various cryo-ET analysis workflows.