Discrete CTF fit resolution jumps

I have a CTF fit resolution plot over all images that seems unusual, and I haven’t an idea how to address it. I don’t know if the data are bad in some particular way, or if the CTF estimation job contained non-ideal fitting values. If in curating the exposures I exclude images beyond 5A resolution, I lose 85% of my data. Has anyone got any insight on this phenomenon?

I think you need to look at the images and think about what went wrong. My naive guess, without seeing any of the actual data, is that all of the higher resolution fit images contain carbon, the sample / ice quality is sufficient for 5-10 Å fits in most images, and >10 Å fits are due to crystalline ice, failed motion correction, or severe radiation damage.

It is also possible the microscope was somehow responsible. Our facility deployed remote collection in response to the pandemic and early on there were a few similar cases caused by forgetting the turbo or confusion with the apertures.

1 Like

I’ve looked at the images and I couldn’t discern from the Thon rings or the images themselves that there was any obvious trend with resolution; crystalline ice appeared relatively infrequently by these observations. But I don’t know how to look for severe radiation damage other than to suspect it as a cause of low resolution. How would failed motion correction manifest? Or, in what part of the patch motion correction job output would this situation be evident?

I had something like this happen- half the micrographs at around 3.5 but then almost half at 25A. For me I think I specified the spherical aberration incorrectly during micrograph input. I think correcting that to 2.7mm fixed the issue, although it may have been that I input the electron dose incorrectly, since I screwed that up as well (put in 1 e/Aˆ2 instead of the 34 e/Aˆ2 that should have been there since I was inputting already-motion-corrected micrographs)

1 Like

@jenchem Dose is only used during motion correction, if you’re importing micrographs instead of movies it doesn’t matter what you enter. (And you have to enter something).

Hi Emily,

what do the CTF fit plots look like for the failed images? If using Patch CTF, with some datasets it tends to find a false optimum with the default defocus search range in the 1D search, and restricting the search range helps - could that be the case here?