Figure 7. Cryo-EM data processing of straight helical polymers using symmetrized refinement yields high-resolution structures.

A. Results of 2D classification after training the filament tracer neural net (step F11). Green boxes indicate classes that were selected for additional processing (step F12) (scale bar, 180 Å). B. 3D volume from the first helical refinement using a featureless cylinder as an initial reference (step F13). The blue arrow indicates α-helices that are present in the flagellin D1 domain. C. A Gaussian distribution of the particle tilt angles indicates proper symmetry was imposed in the 3D reconstruction, further verified by the high-resolution details present in the 3D volume (step F13). D. 3DVA helps to sort particle heterogeneity in the initial particle set. Volume 4 contains 51% of the particles and contains high-resolution features (blue arrow). Volume 1 also displays this quality but contains far fewer particles. The green box indicates Volume 4 was selected for further processing. E. Local resolution map of the final 3D volume along the helical axis F. Top view of the local resolution map of the final 3D volume. The key indicates a map range from 1.75 to 2.75 Å (step F25). G. The tilt angle plot from the final reconstruction displays a Gaussian distribution (step F24). H. The GS-FSC curve of the final map to a global resolution of 2.1 Å. The blue curve indicates the GS-FSC without masking, the green curve indicates the GS-FSC when a soft solvent mask extending 40 Å from the edge of the map is applied, the red curve indicates the GS-FSC when a tight mask extending only 12 Å from the edge of the map is applied, and the purple curve indicates the GS-FSC of the structure with a tight mask and correction by noise substitution. Scale bars for the 3D volumes are 25 Å.