Dear colleagues,
Working with the negative staining SPA, I got a strange FSC of my map after the refinement.
What can be wrong (if anything), and how to fix that?
Thank you!
Sincerely,
Dmitry
Hi @Dmitry,
Thanks for the post; to help us diagnose:
- Which refinement job is this?
- How was particle picking done (which picker was used, or was picking done externally and then imported)?
- When you imported micrographs/particles, what did you set the “Data Sign” or “Negative Stain Data” to?
Best,
Michael
1 Like
Hello Michael @mmclean,
- both non-homogeneous // heterogeneous
- Very interesting question: I picked it in SCIPION and processed later with the cryosparc plugin (in SCIPION). I also have this project run in cryosparc separately. (I need to compare the results).
- Yes, since it was done in SCIPION I probably have to ask the developers. There is no specific button when importing the data for the negative stain samples yet.
At the same time, some other negative staining map finished nicely without the strange FSC. They were also processed in the similar way.
thank you in advance.
Sincerely,
Dmitry
Hi @Dmitry,
I see – and is this a filament? If not, and assuming the particles were imported with their micrograph locations intact, I would suggest first passing the particles through the “Remove duplicate particles” job. This will only keep particles that are adequately spaced apart to prevent duplicate particles from affecting the half-map independence assumption.
If it is a filament, then typically picks are quite close together that if they aren’t sorted into separate half-sets, independence will be violated – to get around this, helical refinement has the option to split particles by micrograph, but unfortunately it’s not implemented in heterogeneous refinement yet. You could still use the remove duplicate particles job if you want to use heterogeneous refinement, but you may lose a substantial portion of your stack.
Best,
Michael
1 Like
Michael, @mmclean thank you!
i will check this and report about the results
No it is not a filament - just a single-particles with C1 symmetry.
Sincerely,
Dmitry
Dear Michael @mmclean,
So, in the beginning, I had 182 000 particles
I removed the duplicates -2000 particles (180 000 remained).
Rerun the non-heterogeneous alignment (with only spherical aberration enabled)
Here are the results.
Interestingly, if you run these (182 000, so with some duplicates) particles in relion – the refinement shows relatively good results, and the FSC curve goes to 0. But then the postprocessing gives quite a problematic FSC curve.
So do you think this result is a normal behavior for this sample&negative staining combo?
(perhaps I mentioned that for some other negative staining samples I got the FSC curve that hits “0”)
Sincerely,
Dmitry