Dataset classifies in Relion, but not in CryoSparc

Hi Folks,

First post here.

I am wondering has anyone ever encountered a dataset which 2D classifies quite nicely in Relion, but not in CryoSparc? If you have, have you discovered any way of making CryoSparc work for such a dataset?

Its actually quite odd, as we can actually classify particles from individual 2D classes from Relion in CryoSparc and get similar looking classes. However if we pool a number of classes or use the entire extracted stack of particles CryoSparc just produces a blob of over-fitted noise. This makes me think that CryoSparc can’t distinguish the real particles from junk.



Hi Thomas,

Try switching the “white noise model” parameter to false. This can make a big difference for crowded micrographs and small particles in my experience.

Also, do you need to use the “ignore first peak of CTF” option in Relion? If so, sometimes just phase flipping in cryosparc seems to work better than full CTF correction.