Question on source files for Ref-based MoCor

Motion Correction
1

Dear all,

I have a somewhat basic question: when Ref-based motion correction is performed does cryosparc go back to the imported .eer files or to the original Motion Correction job after the import. I am asking as both directories (the raw .eer files and the MoCor output) take about the same huge space so it would be great if we can remove safely the raw data to free up space for downstream processing and more datasets.

Thanks in advance!

2

You certainly need the raw movie files and gain reference (.eer and .gain).

I’m not 100% sure whether RBMC reads the rough trajectories precalculated by Patch Motion or not; I don’t think it does… but @hsnyder will know for sure and correct me if I’m wrong.

3

Yes you need the .eer files in order to do RBMC. RBMC also does use the full-frame motion estimates from patch motion correction.

– Harris

1 Like
4

Thanks for the correction. :slight_smile:

5

So, what is going to be pixel size in RBMC? is it original superes pix size or pixel size after motion correction?

6

Please see this explanation.

7

With respect to this specifically, I believe you can remove the unweighted (and if you want, weighted) micrographs from the Patch Motion job manually if you want to reclaim some space. Are you sure though your space estimate is right? The movies are symlinked in the import directory, if the symlinked files are included the storage estimate would be more. I would expect the movies to be larger than the micrographs, but I suppose it depends what the dose rate was for .eer format.

8

based on the reading of explanation, I got the impression that RBMC extract particles from raw image (superres) so we should downsample it if we don’t want super res. I downsampled my raw movie during motion correction 2X. I extracted particles without cropping. now when I am trying to do RBMC with 2 X downsampling, I am getting the homogenous refinement as if it is downsampled 4 times.

Without downsampling during RBMC, I am getting map which is really huge. What is the potential issue and How can I fix it?

Thank you

Dhiraj

9

@dhirajks The “Fourier crop to box size” parameter in RBMC allows you to directly specify the output box size that you want. Your homogeneous reconstruction (after RBMC) should be at that box size. Since you used f-crop 1/2 during patch motion correction, your reference volume is already at the pixel size you want. If you just check the box size of the reference volume and enter that same number in the “Fourier crop to box size” parameter, your output particles will have the right box and pixel size as well.