With the new version of local refinement now taking into consideration symmetry, does this include helical symmetry also?
For example: I have a filamentous particle where the central region is the filament but it is decorated with highly flexible domains coming off, that of course break the symmetry. Using a static mask of the central region in helical refinement gives a reasonable and much better map than doing a global dynamically masked refinement. So I’m wondering if it is possible to do a local refinement with helical symmetry after the static masked helical refinement, as it may slightly improve it? (If not I realize I still have the option to do symmetry expansion…etc)
Thanks and best wishes,
Right now, local refinement only supports helical symmetry by way of symmetry expansion. It sounds that this could be helpful in your case – after doing the helical refinement, you can read out the final twist, rise, and helical symmetry order parameters from the end of the streamlog, and run a symmetry expansion with those values, and then use those for a future local refinement. You should make sure in the parent helical refinement that the “Limit shifts along the helical axis” is on. The main downside of symmetry expansion is only that it makes your particle stack larger, hence refinements will take longer.
In the specific case of helical symmetry, if you suspect that there’s still flexibility in the central core, you can use a mask that covers fewer asymmetric units. You can do this in the volume tools job – pass in your static mask, and set the “Z-clip fraction (for masks)” parameter to 0.5 for a mask covering 50% of the box size. Also make sure to set all of the other mask parameters (i.e. set the threshold to 1, and use the same dilation radius, padding width, etc. that you used to create the static mask). It’s a tradeoff: with longer masks, flexibility will generate uncertainty around the maximum likelihood alignments, but with shorter masks, you’ll have less signal available for alignment… so more uncertainty again.
There was a recent similar post where I added some other notes about this workflow, but the only catch is that you should make sure that the “Force re-do GS split” parameter is off in the local refinement with expanded particles – this should preserve gold standard independence assumptions.
Thank you Michael. Very helpful.
We’ve just updated to 3.1 last night and I’m giving the new helical refinement parameters a try. So good so far for a test set where I get the same resolution as before in 3.0 (its not an ideal sample but it goes to 3.7A). I’m just about to try helical expansion but I am wondering: My helical unit is actually a dimer so I impose D1 symmetry during helical refinement as helps a lot. When I do symmetry expansion do I put in D1 or C2 as the desired symmetry? I also assume that it must be C2 symmetry when I attempt local refinement as I’m interested in the dimer units?
I would think that if you did a helical refinement with D1 symmetry, you should symmetry expand with the same D1 + helical symmetry parameters as the helical refinement, and then use the symmetry expanded particles for a C1 local refinement. D1 and C2 are both 180º rotational symmetries around an axis, but D1 is around the dyad axis (y in cryoSPARC), whereas C2 is around the z axis, so they aren’t the same.