Convert csv file to star file and keep orientation parameters

Dear all,

I was wondering if there is any established way to convert the csv file after refinement with cryosparc back to star file and keep the angles and translation parameters of each particle? In such a way that you could for example just run 2D or 3D classification without alignment within RELION

Thanks a lot in advance!

Best wishes,

Rafael

@Rafael-Ayala you can use csparc2star.py in Daniel Asarnow’s pyem package for this (https://github.com/asarnow/pyem). That said, I have tried this but haven’t had much luck continuing refinement in relion, for reasons that are not entirely clear to me, so I’d be interested to hear if you have a different experience.
Cheers
Oli

1 Like

@olibclarke thanks a lot and sorry for the late reply! I have been testing the script, and got strange results

I have converted a set of particles from csv file back to star file and tried the following with RELION:

  • Displaying particles seems to be alright (nothing strange at first sight)
  • 2D classification with alignment: seems to work fine, although surprisingly the resulting classes seem to have a different contrast/birghtness (the white intensity of the resulting images seems very high when displaying at a sigma of 3). Interestingly, if I display the particles corresponding to the classes they look normal
  • 2D classification without alignment: simmilar problem with the contrast of the classes
  • 3D classification with alignment into 1 model (I know this set of particles is homogeneous): comes up with a very strange elongated model
  • 3D classification without alignment into 1 model: even more elongated model

I was wondering if you had found similar problems?

Best wishes,

Rafa

@Rafael-Ayala I must admit I haven’t investigated it in as much depth… I just found that continuing refinement from a star file generated by this script in RELION was unsuccessful no matter what parameters I used, whether local or global searches the estimated accuracy of angles remained very high and the map lacked details. When I get a chance I will test it more thoroughly, but there would definitely seem to be something funny going on.

Cheers
Oli

I have updated the tutorial on the pyem github wiki. The most recent program version can also still be found on in the github repo.

@Rafael-Ayala @olibclarke I am sorry you had issues with my program. I am quite sure that csparc2star.py converts pose representations correctly, as I frequently reconstruct densities using converted alignment parameters.

I would need more specific information to understand what went wrong exactly in your hands. I’d appreciate it if you also took the time to consider the suggestions in the last section of the tutorial.

Hi @DanielAsarnow, actually I have been using your program very frequently to go from cryosparc to RELION and have been getting excellent results always! I am not sure what was going wrong back then, but seems to work all fine since a long time ago!

Thanks a lot!

Best wishes,

Rafa

Likewise - has been working fine for me for a while!

Cheers
Oli