I am working on a D2 symmetric 55 kDa protein complex. What is your take on explaining a resolution drop between NU refinement with D2, marginalization on - 3.7 Å vs. symmetry expanded (D2), and local refinement over the entire complex - 3.2 Å? Does this indicate pseudo-D2 symmetry?
local refinement with mask on everything? or just one monomer?
If it is local refinement with a mask over the entire complex then I don’t know exactly how to explain this - it should be pretty equivalent to enforcing symmetry in global refinement (perhaps some difference in settings is responsible). If it is improved with a mask over an asymmetric subregion then that likely indicates intersubunit flexibility within the oligomer.
If you want to test for pseudosymmetry, you can try either symmetry relaxation in NU-refine, or classification without alignments after D2-refinement (with a mask around the entire complex).
Thanks for helping: The mask for local refinement is on the entire complex. To be more precise, Local Ref after sym expansion was run in C1.
1 Like
Yeah - I wouldn’t say this indicates pseudosymmetry (because you are starting from something that had D2-sym enforced and refining with local searches for the entire oligomer). Not sure exactly why the improvement then…
If you’re symmetry expanding and refining over the entire complex aren’t you increasing your particle stack size by 4? That would almost certainly improve your resolution.
Yes but in C1, vs 1/4 the particles in D2 (4-fold averaged via symmetry) - should come out to same outcome I would have thought?
Good point. Could be that the restricted search ranges lead to better particle alignments. Is there any evidence for large rotations/translations during local refinement suggesting a difficult to align subset of particles? If the particles are already pretty well defined I would be curious what the results of subsequent NU refinements with only low pass filtering the starting reference to 4-6 A rather than the default 12.
Good point, looking for larger shift changes. Here are the logged changes and job parameters for the D2 expanded local Ref. The initial lowpass in the D2 (relaxed, maximization) was set to 6 A. No improvement over D2 expanded local Ref.
Are those shifts from the final iteration or first iteration? With recenter shifts and rotations turned on you won’t be able to tell how much a given particles changes over the full job, just how changes occur for each iteration. You can rerun with re-centering turned off to get a better idea for how your entire particle stack is changing over the entire job. Runing with re-centering is a good sanity check though to make sure that alignments are generally improving each iteration.
