Good particles (from cryosparc) import into Relion result in bad 2D classification results


I want to run a polishing job to improve resolution. First, I imported my good particles from Cryosparc V4.2 into Relion3.1 by using with a inverty option. Then I re-extract these particles in Relion3.1. However, the 2D result in relion was very bad with only one class while in Cryosparc was good( See below picture).
2D result in relion:

2D result in cryoSparc:

I have done this successfully in several other projects, but I failed in this project and I didn’t know why. By the way, micrographs in this projects were collected in a different facility. So, what’s the problem here?


Have you tried the conversion without the inverty flag?


1 Like

In Relion, under the CTF tab, there is an option called “Ignore CTFs until first peak?”, try setting this to “yes” and repeating the 2D classification.

the “do I invertY” conversation is alive and well, go re-check that thread.

Hi Yang,

I have tried the conversion without the inverty flag,the 2D result in relion was worse(see picture below).


Hi Sia,

I have also tried the option you mentioned in relion ( Ignore CTFs until first peak----yes). The 2D result seems better (the first two classes) but still weird (left classes). And I have imported about 20k good particles into relion, while the first two classes (good particles in relion) only contained 8k particles.

You’ll need to tighten the mask in RELION - it’s locking on and aligning the strong densities at the edges which are not centred, resulting in the odd 2D.

You could also try running 2D classification without alignment and see what it output (if the alignments from CryoSPARC are held over…)

edit: Also, due to differences in the algorithms and how RELION and CryoSPARC do things, classifications will look a bit different.


Thank you for your suggestions. I have tried both the option with a more tight mask and the option without alignment in relion 2D classification. Both of them produced a better 2D result.

Thank you very much.