Manually deleting .mrc files from output of Patch Motion Correction

Motion Correction
1

Hi,

Among 3 .mrc files (basename_patch_aligned.mrc, basename_patch_aligned_doseweighted.mrc, basename_background.mrc), would it be okay to manually remove 2 .mrc files other than a doseweighted one from the Patch Motion Correction job folder inside the project directory?
Mostly for sake of saving disk space up.

Related, I wonder if such a maneuver would potentially cause a problem down the pipeline should one wish to run Reference Based Motion Correction with the Patch Motion Correction job as an input.

Best,
Jinsung

2

I’ve deleted *patch_aligned.mrc micrographs without ill effects on many projects to clear some space up, but you won’t be able to redo Patch CTF Estimation if you do. It doesn’t effect Local CTF Refinement. Just make sure you’re happy with your base CTF estimates first. :wink:

3

If you are in a pinch and do not want to re-run Patch Motion, you could soft-link _patch_aligned.mrc to point to the equivalent _patch_aligned_doseweighted.mrc. The Thon rings might be slightly weaker in theory, but probably it would be just fine.

Here is a bash “one-liner” to do so:

for FILE in *_patch_aligned_doseweighted.mrc; do
ln -s ${FILE/_doseweighted.mrc/mrc} $FILE
done

(Not tested, so do a test run in a mock directory first!)

4

Oh, good point, I forgot to mention I’ve tried this (only once, though)…

Patch CTF on doseweighted micrographs gave very poor CTF estimates. Might have been dataset dependent, but I went from ~95% of micrographs with good CTF fits and estimating <4Ang to >50% having fits worse than 15Ang.

Would be interested in further feedback if anyone else tries it.

5

Interesting that you got that poor results from Patch CTF on dose-weighted micrographs. I have not tried in CryoSPARC, but did some testing long time ago using the RELION workflow with MotionCor2 and Gctf and as far as I remembered, it wasn’t that big of a difference.

6

I didn’t expect it either.

I’ve tried dose/not-dose weighted in both RELION and CryoSPARC, with CTFFIND it seems to make little difference (in either suite) but with Patch CTF estimation, using dose weighted output seemed to really hurt.