Maximum likelihood refinement of electron microscopy data with normalization errors.

Area of Science:

• Structural biology
• Biophysics
• Computational biology

Background:

• Standard data models for cryo-electron microscopy (cryo-EM) refinement are sensitive to normalization errors.
• Variations in background, signal brightness, and image artifacts are common in cryo-EM data, hindering accurate classification.
• Existing methods struggle with structurally heterogeneous datasets.

Purpose of the Study:

• To develop a robust statistical data model for maximum likelihood refinement that accounts for normalization errors.
• To derive an algorithm for maximum likelihood classification of heterogeneous cryo-EM projection data using the new model.
• To demonstrate the utility of the enhanced model on challenging biological datasets.

Main Methods:

• Development of an extended statistical data model incorporating normalization error parameters.
• Derivation of a maximum likelihood classification algorithm based on the new data model.
• Application and validation of the algorithm on cryo-EM datasets of 70S E.coli ribosomes and human RNA polymerase II.

Main Results:

• The novel data model effectively handles common normalization errors in cryo-EM data.
• Maximum likelihood classification using the conventional model failed for the tested heterogeneous datasets.
• The new approach successfully resolved previously unobserved conformations in both ribosome and polymerase II complexes.

Conclusions:

• The proposed statistical data model significantly improves the classification of structurally heterogeneous cryo-EM data.
• This approach overcomes limitations of conventional models in the presence of normalization errors.
• The generalized algorithm has broad applicability for other maximum likelihood methods in cryo-EM research.