I have a set of symmetry expanded particles (symmetry: T), and I want to further classify them before I run local refinement. I tried 3D classification with a mask on a single particle, but the final map after local refinement looks not very good, as the resolution is definitely not as high as it says. I’m considering maybe the particle itself are somewhat dynamic, and a classification with alignment can help me classify them better, but I saw that heterogeneous refinement is not a good option for symmetry expanded dataset. What job or parameters can I try?
Is there any particular reason for avoiding Local Refinement before classification? It would be better to confirm that there are particles with sufficiently correct angular assignments before proceeding with classification.
did you run NU-refine (or at least homogeneous reconstruction, also available to do for all classes in a single job) following 3D class? the volume outputs of 3D class are limited to the resolution you provide, and the local refinement could be more tricky to achieve high resolution model
Thanks for the advice! My concern was that after symmetry expansion, there were 4 million particles, and I thought using a sub-class of them for local refinement would be ideal to save some time.
I will try running local refinement first. If I understand it correctly, when I use the output particles from local refinement job for 3D classification, the alignment information would be included, so I don’t need to worry about the alignment during 3D classification?
On a second thought, since my 3D classification is on symmetry expanded data, is it OK to run NU-refinement following it? Maybe I should run Remove Duplicate Particles first?
