Question about MotionCor2

Hi
I have some questions about the FmDose and the results of the MotionCor2.

  1. I wonder when the FmDose value will be modified to frame dose instead of total dose.

  2. When I checked motion corrected micrographs, It looked the same as before the gain reference was applied. I turned the gain reference 90 degrees when import movies and MotionCor2 command line turn gain reference 90 degrees again. I am wordering if the rotation is applied repeatedly.

  3. MotionCor2 seemed to cause errors that stopped often due to the GPU. If works stop because of this reason, is there any way we can continue from there? I’d like to be able to continue because the time is too inefficient to start all over again, and I’m not sure there’s no error in the middle again.

Thank you :smile:

@KSJ Motioncor2 has multiple gain rotation arguments. There is -FlipGain for which 0 does nothing, 1 flips upside down, and 2 flips left-right, as well as -RotateGain which allows for rotating by 90, 180 or 270 degrees. The common situation is needing -FlipGain 1 when a .dm4 gain from DigitalMicrograph is converted to .mrc (e.g. e2proc2d.py SuperCDS.dm4 SuperCDS.mrc).

You can inspect the uncorrected gain artifacts caused by wrong rotation/flipping to figure out which argument is needed. If it’s an upside down flip, there will be a dark spot somewhere and a bright spot in the same X position but the opposite Y.

What are your errors like? Generally it’s recommended to pass -GpuMemUsage 0.5 to avoid any GPU memory pressure issues. Also, the system memory requirements are fairly high. For K3 images with optimal collection parameters (~8 e/px/s, ~frame time 0.05, ~8 s exposure, ~100 - 200 frames) you should save around 38GB per Motioncor2 process.

1 Like

Thank you for kind answer @DanielAsarnow!

  1. Is there a way to set -RotGain in cryosparc motioncor2 beta version?
    I don’t know how to prevent multiple gain rotation because the rotation value set at the ‘import movies’ step seems to go in the motioncor2 automatically.

  2. I practice cryosparc with EMPIAR data and it’s size is quite big (over 900GB).
    Error code is this.

File “cryosparc2_worker/cryosparc2_compute/run.py”, line 82, in cryosparc2_compute.run.main
File “cryosparc2_compute/jobs/motioncorrection/run_motioncor2.py”, line 364, in run_motioncor2_wrapper
_,output_micrograph = mrc.read_mrc(output_path_abs)
File “cryosparc2_compute/blobio/mrc.py”, line 105, in read_mrc
with open(fname, ‘rb’) as file_obj:
IOError: [Errno 2] No such file or directory: u’/media/Data/ksj/cryosparc/nanodisc/P4/J187/motioncorrected/LRRC8A_ND_1571_motioncor2_aligned.mrc’

When I did it again, there was no problem. So I think the file is fine but GPU is the problem.

Now I’m splitting up the input files.
But in this situation, instead of starting from the beginning or splitting files, I want to continue. Is there any way?

IOError: [Errno 2] No such file or directory: u’/media/Data/ksj/cryosparc/nanodisc/P4/J187/motioncorrected/LRRC8A_ND_1571_motioncor2_aligned.mrc’

There’s your problem.

Thank you for answer!

But it went well when I repeated it again with same condition :cry:

Instead, the next time I did the same thing, the same kind of error occurred in another file and that file also went well when I did it again.

So I didn’t think it was my problem, but is there a problem with the file or me?

Seems like a filesystem issue. Maybe your drive is failing, or going to sleep too aggressively. Is it an external or network drive?

It’s external drive!

First thing is try a different USB cable, then try a different drive. Take some of these images that fail sometimes and run the job several times.

For the gain, just use whatever setting in import makes the images already lined up with the gain you have, and don’t rotate it any more.

Thank you @DanielAsarnow ! I’ll try!

HI @KSJ,

This bug is fixed in the latest version of cryoSPARC, v2.15.0