Dear all,
Recently I have collected data on 2nm carbon coated grids for an asymmetric 200kDa long protein. The power spectrum of those movies I am attaching below. It looks weird for me. Can someone explain why it is so?
your pixel size is incorrect. check the “upsampling” and the super-resolution.
2 Likes
I want to second @CryoEM2 and say that your import job is a likely culprit.
If you collected in super resolution be sure that’s reflected in that job parameter (you can always crop by half in patch motion anyway).
Mistakes of this nature are evident when the power spectra appears to have “tears” or strange patterns of white and black spots.
1 Like
I’m not discounting this as being correct, but I’ve never seen this result from - even deliberately when I was demonstrating a point to a training course - getting the pixel size wrong. Although I’ve never been out by an order of magnitude or similar.
I have, however, seen CTF misfits like the second image when the default patch CTF parameters are not loose enough; e.g. with a very uneven grid and a fairly high defocus range having some micrographs exceed the 4um default maximum, which gets the patch algorithm really confused.
Could u please state clearly. what I should try now to troubleshoot?
Pixel size has given correct only.
my assumption is that your microscope was set to operate in counted-super resolution mode, and the “configuration” tab of cryosparc live used a pixel size that is normal for counted mode (but not super resolution). Try using a pixel size exactly half as large (probably ~0.5-0.6). you will also need to change “upsampling” to 2, or extract particles in a much larger box size (since box is in pixels).
Yeh, the pixel size was half only. Up sampling is 2.
I dont understand what I am missing
Is it fine if i bypass the CTF values (second image) and consider them as good enough to be taken to next stage.
Assuming the pixel size is correctly input (remember, put the “4K” pixel size in at import for EER files, even if you want to work with 8K micrographs, CryoSPARC recalculates automatically!) re-run patch CTF, set the maximum search defocus to something like 60000 and see if the fit is better.
If you didn’t enable 16-bit MRC, you can also try CTFFIND.
edit: to directly address your question, no, I do not recommend ignoring the incorrect CTF fitting.
I think this is irrelevant given that EER data is inherently 4K, 8K and 16K all at the same time. CryoSPARC always wants the 4K pixel size at import and will recalculate final pixel size based on EER upsampling value as necessary.
Am I reading this correctly?
You’re suggesting telling CryoSPARC the “super resolution” pixel size (which is not the right way to handle EER data import), on top of enabling EER upsampling of 2?
Or I’ve got a stinking cold and I’m misunderstanding your recommendation.
Please confirm:
- type of data (EER from screenshot)?
- 4K pixel size (magnification and whether filtered/unfiltered Falcon 4(i) perhaps also)
- what parameters were set for:
– import,
– patch motion and
– patch CTF (if not defaults)
The local motion trajectory plot looks a bit suspicious; CryoSPARC usually only uses so many knots at fairly low magnifications.
Bad airport advice. Do as rbs_sci says, they’ll guide us out of this mess 
1 Like
EER data
15000kx magnification
Falcon 4 data
Import (pixel size - 1.9, abberation - 2.7. voltage -200, electron dose - 60
patch ctf and motion correction has been done with default parameters
OK, 1.9 Ang is pretty low mag, so the number of knots makes sense. That solves one curiosity!
You left EER upsampling at 2, which would mean 8K output and effective 0.95 Angpix.
I’d do another Patch CTF and increase max defocus as previously suggested, could also try setting number of knots to 1 and 1. Or try CTFFIND since you saved in 32-bit MRC (which is the default).
You’ve got nice power spectra, it’s just finding how to optimise parameters which are obviously outside of the defaults. 
sure. will try my best 
1 Like