Converting from Cryosparc to Relion using csparc2star.py: Box size error

Hi all (@DanielAsarnow),

Have just spent a while finding my way around csparc2star.py. It was having trouble converting the output of a cryoSPARC refine job to Relion star file (wanted to put my model through Bayesian polishing). I followed the instructions on https://github.com/asarnow/pyem/wiki/Export-from-cryoSPARC-v2 . However, Relion continued to throw the error: ERROR:
ObservationModel::getBoxSize: box sizes not available.

I have solved the issue (I think, job just running so we’ll see how it looks, but no error so far), but to do so I had to manually add rlnImageSize to the .star file header, despite using the --boxsize kwarg to specificy the box size in the original convert command input.

Just wondered if this was a quirk you were aware of, and whether it might be worth adding something into the above tutorial about checking that .star files contain all the relevant information. I am relatively new to the EM data processing so maybe this is something that would be obvious to more experienced users, but I just thought I would mention it.

Cheers,
Matt

--boxsize is to account for using a different reconstruction size in cryoSPARC.

Since Relion won’t check the images outside of Import jobs, you must add the particle size yourself like you discovered. Or you can use --relion2 and Import the resulting star file in your Relion project. I think editing the .star is a little less work.

I found this page having encountered the same confusion as @mattarnoldbio, following that otherwise very helpful pyem/wiki page. Flagging the need to add rlnImageSize manually would certainly have been a welcome tip.