I was able to get cryosparc 2D classes with secondary structure present in the classes. However, when I transferred the particles to Relion (excluding the low confidence particles with the cspar2star program @DanielAsarnow 's very helpful program and trying with and without image alignment in Relion), the classes I received were slightly blurry versions of the ones I got in cryosparc. I was wondering if anyone had any insight on how to get the same quality of 2D classes in Relion as you do in cryosparc?
You might try using the “ignore CTF until first peak” option - for some datasets in Relion this can make a big difference, if the low frequency parts of the CTF are not well modeled.
This actually did help a bit when used thank you! Just wondering did you have any suggestion about image angling and alignment settings?
@adtaheri How do the cryoSPARC classes look if you change the resolution limit from 6A (default) to 12A? Or if you use marginalization instead of max over poses?
Perhaps these features are actually due to a bit of over-refinement.
I actually always run using 10A for this reason. I’m actually starting to see some of the secondary structure in relion as I did in cryosparc, just not as clear. I’m wondering if I should expect to see similar classes in relion and cryosparc, or simply get them to the best quality as possible and move on?
I’m sure they can be made similar with further parameter optimization. For me, I usually “reject bad” 2D classes rather than “accept good” so I would do that and move on.
If you need to “accept good” (which I don’t recommend due to the rare view problem), then it’s probably worth continuing to test other parameters.