Crsparc to relion stack file

hi, all:
In cryosparc v2, I manual and template picked the particles, after 2D classification, I got a set of “good” particles, now I want try this set of particles in relion, I used cspar2star to get a star file, but how can I get stack file(.mrcs) and which file in cryosparc to convert? can anybody tell me how to do it?

thanks a lot!


1 Like

The pass_through files contains the location of particles, and what you need to do is to let relion know where they are. You can make a symbol link to relion directory. Also, the extensions of particle names (.mrc) should be changed to .mrcs for relion recognize, both in symbol link name and star files.
That could work. But last time I do it, relion itself got some reading mrc files problems which make relion failed to do Initial model building. I reinstall relion to overcome it.

Thanks a lot. but I still not fully understand, because all my data is in cryosparc, from GitHub “Export particles from cryoSPARC” example: --data-path Import/job005/original_particle_stack.mrcs …
I try it , but not working, it returns:
‘ error: unrecognized arguments: --data-path’

can somebody help to explain how to use this command,




I just convert the .cs files to .star files using like this: cryosparc_P7_J35_005_particles.cs
And use the star files to do relion works.


I mean if you check the particles names in .star file you can find it like this way: imagename(000001)@relative_path. What you can do is let relion konw that relative path. In my way, I just created a symbol link of the particles directory to relion’s directory.

thanks a lot. I will try it.

If you merge particle coordinates back in with --copy-micrograph-coordinates you will get the original image paths by default. The coordinates are very often useful so I would recommend that way. You could also use a simple find-and-replace, and @Y_Z has a good solution as well.

If you really want a single stack, you can use or relion_stack_create to generate a stack from your converted star file. I’d recommend (of course, I wrote it) because it only needs ~1 particle of memory at a time instead of the entire stack.