Comparing Patch (or Full frame) Motion and MotionCor2 corrected micrographs

Motion Correction
#1

This may not be an issue, just trying to understand the difference between Patch (or Full-frame) Motion and MotionCor2, see below the same movie was motion-corrected by Patch Motion (top) or MotionCor2 (bottom), the top mic looks quite uniform, while for the bottom mic, left side is darker than the right side.

(Top)
image

(Bottom)
image

Also, the same particle set extracted from MotionCor2 corrected mics leads to ~0.3 Å lower resolution with Non-uniform refinement (than that from Patch Motion corrected mics), even after RELION Bayesian polishing.

Thanks in advance for your comments.

#2

Hi @bing,

I would suggest you double check your gain reference - I suspect it may need flipping for motioncor2 (or not flipping, if you have flipped it).

Cheers
Oli

#3

Thanks for reminding @olibclarke!

The movie is in TIF format, so I did Y-flip the gain reference for Patch Motion Correction, but no flipping for MotionCor2 (during the movie import step). I consulted with our EM manager, he suspected this was due to different algorithms that Patch Motion Correction and MotionCor2 used, perhaps Patch Motion Correction auto-scales/normalizes the contrast locally, while MotionCor2 doesn’t.

#4

Even with the right gain reference applied I would expect the patch motion corrected data to give better resolution.
This is at least what I see regularly.

Best,
Tarek

#5

Hi @bing,
Like Oli said, it looks like related to the gain reference. What happens if you run stand-alone Motioncor2? Did you try all possible orientations (flip and/or rotate the gain reference)?
Also, (you probably noticed already but in case…) as the two micrographs you provided are mirror images of each other over the x axis, did you take that into account to compare the “same particle set” ?

Best,

#6

@lezark Yes MotionCor2 corrected mic is Y-axis flipped relative to Patch Motion corrected mic. I did take that into account to gain reference or particle set conversion.

I somewhat believe the uneven contrast is due to the variable ice thickness of the imaging area. and Patch Motion (or legacy Full-Frame Motion) can normalize the contrast so that the whole micrograph looks uniform, but not for MotionCor2 or Unblur.