I have a dataset that is mostly helical, but there’s a subset of particles that have 0 pitch. In this case, i thought i could just refined it with normal non-uniform refinement with Cn symmetry enforced. What i realized after running the job is that the FSC got elevated probably because the shifts was too much so that i got many duplicated particles. I could just remove duplicated particles and continue with local refinement, but that would remove some good particles.
Is it possible to add local search capabilities in normal refinement? If not, is there a good way to deal with my problem in cryosparc?
Can’t you just start from local refinement with Cn symmetry? What advantage does “normal” refinement offer if you are limiting searches? Or is it that you want global search of orientations, but limited translational search?
Thanks for the reply! The problem is that i don’t have initial angles to start local refinement right away. If i start a global refinement to get the angles then the shifts got messed up.
Ah I see… so you effectively want to do a global orientation search but limiting shifts along the helical axis specifically? I don’t think there is a way to do that as such in CS. There is an option to limit shifts along the helical axis in helical refinement - perhaps if you set the rise/twist appropriately you could use that to get initial angles, then proceed to local?
Yes exactly. Doing a helical refinement with very small rise is a good idea. I’ll see if that works. Thanks a lot!
Edit: Also want to add that this is not a perfect solution either. Because if one set the rise to be very small, the “limit shifts along helical axis” parameter would basically lock the y shift, which is not ideal. Then if I turn off “limit shifts along helical axis” the same problem as global refinement occurs. It would still be the best if i can manually set the limit of the shift search.