Particle migration analysis in iterative classification of cryo-EM single-particle data

Area of Science:

• Structural Biology
• Biophysics
• Computational Biology

Background:

• Heterogeneous samples in cryo-electron microscopy (cryo-EM) present challenges for accurate biological structure classification.
• Current classification methods can be subjective, lacking robust statistical grounding in cryo-EM data.

Purpose of the Study:

• To develop a quantitative method for determining classification convergence and the number of distinct biological classes from cryo-EM data.
• To reduce subjectivity in single-particle analysis by basing decisions on data statistics.

Main Methods:

• Proposed a novel 'jumper analysis' technique for iterative classification schemes.
• Initiated classification with an excess number of classes, then combined similar reconstructions.
• Identified similar classes by analyzing particle migration ('jumpers') between iterations post-convergence.

Main Results:

• Demonstrated jumper analysis on a benchmark cryo-EM dataset using RELION 3D classification.
• Successfully identified classes yielding similar reconstructions based on particle movement patterns.
• Provided a quantitative basis for determining convergence and class number.

Conclusions:

• Jumper analysis offers a statistically robust approach to cryo-EM data classification.
• This method contributes to reducing subjectivity and advancing automated single-particle reconstruction.
• Represents a significant step towards fully automated cryo-EM data processing.