Flipping and rotating gain reference

Import and Export
1

Dear cryoSparc team and users,
I have an unusual question. We know from our data collection, that the gain reference is flipped in X and rotated 180°. We gave these settings whenever we processed in Relion, as it was needed. Recently, someone told us that we should select these setting in cryosparc, because it does so automatically. Is it true? Does cryosparc handles .tiff and .mrc files differently during import? I don’t think there is any mention of this in the official guides.

Many thanks.

2

Hi,

AFAIK, RELION flips TIFF images along the slow axis (i.e. flip in y-axis). In contrast, cryoSPARC reads TIFF data in the order it’s written on disk (i.e. does not flip in y-axis). If your data (e.g. .tiff stacks, .mrc gain reference) required a flip in y transform in RELION, then this is not required in cryoSPARC due to the aforementioned difference.

Whenever in doubt, you can try it both ways and inspect the resulting thumbnail preview. It is often fairly obvious.

Cheers,
Yang

PS, If, however, you plan to use the MotionCor2 wrapper from within cryoSPARC, then a flip may again be required. MotionCor2 and RELION share the same TIFF reading convention.

2 Likes
3

Thank you, this was very helpful!

4

It is just an extension of the conversation.

  1. Does it always necessary to flip gain reference (GR) file along Y axis wile importing, if our movies are in tiff format.
  2. I have two Antigen-Antibody complex structures solved around 3.5A. But I did not flipped GR while importing the tiff movies.
  3. What is the reason of flipping or non flipping of the GR file?

Thanks

5

@leetleyang Is correct above - the y-flip option will be switched between Relion/WARP/MotionCor3 and CS for that reason.

@Shahsal The reason for additional rotation/flipping the other axis or not is literally that the camera is installed upside-down or sidewise. The K3 in particular can usually fit in the microscope in 3, if not 4, different orientations and there is no guarantee the engineer has done so “correctly” - with the tilt axis roughly parallel to the long axis of the K3 (tilt axis rotation ~85˚ or ~175˚ as calibrated by SerialEM).

1 Like
6

@DanielAsarnow, that is a really interesting fact I did not know :). Why is the gain oriented differently relative to the movies? Does it turn one but not the other?

7

@frozenfas file formats as mentioned above. Convert a super res .dm4 reference to .mrc yourself (e.g. with e2proc2d.py) and compare to the .mrc counting gains output by DM.