Hole Edge in image leads to incorrect CTF Fit estimate

With CSLive, having a bit of hole edge in image sometimes leads to an incorrect estimate of CTF Fit. The Thon rings and 1D CTF fit to the data is correct, except that the fit drops off twice - at 25A and then at the high resolution cut off for the data. This is most frequent with gold grids.
Sorting by CTF fit is quite convenient. Is there a work around, or, perhaps, a way to Accept an image that was rejected based on cutoff?
Thank you.

GoldEdge_1DCTF GoldEdge_Image !

If you need to shoot near the edges on holey gold, then my advice to set the low resolution limit for CTF estimation to something higher, like 20 A. I think even higher will still converge well, but in your example 20A would be sufficient because cryoSPARC uses a 0.3 cut-off for determining the “fit resolution.”

Also, if you mean to shoot the hole centers, and still have lots of edges, then the image shift calibrations or alignment between magnifications are wrong. Of course, at lower mags, hole edges are hard to avoid. I’ve found that on holey gold 1) objective aperture is absolutely required, even with the energy filter, 2) energy filter helps a lot, and 3) 300 kV is much better than 200 kV.

With 300 kV, 100 um objective aperture, GIF, and the CTF trick above I was able to get to ~3.0 A from intentionally shooting hole edges on holey gold, so it’s possible. Good luck!

Further to Daniel’s point above, this is something that we’re aware of and actively working on. A solution for this will be coming out fairly soon, so there will be no need to change the default job parameters.


Thanks everyone!
@DanielAsarnow, thank you for your recommendations. For most of our collections, edges are the result of trying to cram as many shots per hole as possible. Since there is little direct control for targeting location of exposures in SerialEM, a bit of an edge is the cost of having an extra shot or two.

Do you use “regular pattern”? You can use “custom pattern” instead if you want something specific. You can even shift the image to the desired location in record mode, then save the shifts, to make the pattern super accurate. IMO it’s not worth the trouble, though, since patch CTF works fine anyway.