Export particles from cryosparc to relion for Bayesian polishing?

Hello, I am trying to export a particle stack from cryosparc into relion so that I can perform Bayesian polishing in relion.

I used csparc2star to convert the particle stack from cryosparc to relion, then modified the path for the motioncorrected micrographs so that it points to the relion motion correction and not the cryosparc motion correction.

As a check, I attempted to re-extract the particles from the relion micrographs, and everything ran fine until the very end of the extraction, as shown in the output log:

=== RELION MPI setup ===

  • Number of MPI processes = 4
  • Leader (0) runs on host = cd-grout.mit.edu
  • Follower 1 runs on host = cd-grout.mit.edu
  • Follower 2 runs on host = cd-grout.mit.edu
  • Follower 3 runs on host = cd-grout.mit.edu
    =================
  • Re-extracting particles based on coordinates from input _data.star file
  • Import/job014/a2only_for_postproc_swapxy_2.star
    Extracting particles from the micrographs …
    1.70/1.70 hrs …~~(,_,">

Joining metadata of all particles from 21653 micrographs in one STAR file…

RELION version: 4.0-beta-1
exiting with an error …

The error message was:

ERROR:
ERROR: optics group “” does not exist in micrograph STAR file…


MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD
with errorcode 1.

I think this has to do with the fact that even though I changed the motion correction path to the relion micrographs, I haven’t changed the star file parameters associated with the relion-processed micrographs (i.e., the values for rlnOpticsGroupName, rlnCtfBfactor, rlnDefocusU, rlnDefocusV, etc.). Can anyone tell me how to update the star file from cryosparc with these values from the relion-processed mics?

Additionally, can anyone tell me if I’m amiss in my approach here? I’d like to take these particles and, in relion, do a quick round of 3D classification, then 3D auto-refine, then per-particle CTF estimation and Bayesian polishing, so I need to make sure that the star file generated from the cryosparc particle stack works with the relion-processed micrographs.

Hi,

See here: Particle Import to Relion from cryoSPARC error- optics group "" does not exist in micrograph STAR file - #2 by ajian

Thank you for the reference, but I’m still a bit confused. In the example in that thread, it looks like the optics group is already present in the star file (as “1”), so I’m not sure what I’m supposed to change.

Secondly, I have already divided these micrographs into optics groups in relion because I collected them using AFIS. Is there a way to update the optics group numbers in the cryosparc particle star file to match the AFIS grouping?

Ah, sorry, it’s still early here - I see now how the previous thread fixed this issue, but I’m still not sure about how to update with the AFIS grouping. Any clarity on that point would be much appreciated!

You can do this using Exposure Group Utilities - I haven’t done this for AFIS groups, but others have and apparently it works, though seems kind of complicated: Importing Optic Groups from EPU AFIS

I’m not sure if I’m understanding the issue without looking at the data, but this sounds like the problem that occurs when the optics group numbers in the optics table do not match the optics group numbering in the particles table or if there is an difference in the optics group numbering between the particle star file you’re extracting and the micrographs.star you provided as the input.

Sometimes the optics groups in the particle table will all be changed to 1 and the easiest solution is to split starfile into the optics groups (this is easy to do for serialEM data with the shift position in the micrograph name but I don’t know if it’s the same for EPU?). For each optics group star file you can then change the group number to 1 in the optics table and when you join them back together they will all be renumbered as long as the group name is unique.