CryoSPARC-5.0.6 vs CryoSPARC 4.6.2 resolution changes for non uniform refinement

3D Refinement
1

Hi,

I am trying to determine a structure of a 50kDa protein bound to a 25kDa protein complex. with my current data I was able to get good 2D classes. And using those particles I carried out ab initio into 3 classes, and chose the best class for non uniform refinement and the best resolution I got was around 4.99 A (and upto 5.2A). When i looked at the density and docked my starting models i was able to place my two protein submits with a decent amount of confidence, with lower resolved regions in the 25kDa protein parts, but i was able to place 50kDa protein pretty confidently.

For that specific NU refinement job I used CryoSPARC 4.6.2, mostly default parameters with the following parameters changed:
Initial low pass resolution (A): 25
Dynamic mask far multiplier (A): 10

Somewhere between then and now our cluster went through a software update to CryoSPARC-5.0.6, and I am getting far lower resolutions (ranging between 7 to 9A) for the same job using same data. And I am trying to figure out what parameters can be tweaked to get it better again (or make it behave as close to the previous version). Or any other recommendations to get a resolution in the range of 3 to 4A.

That is the short summary of my current issue. Pleaselet me know how to get around this issue.

Thanks

2

You can try the parameters outlined in this paper, which could be quite slow.

https://www.biorxiv.org/content/10.1101/2025.09.08.674935v1

The new job type
Homogeneous Ab-Initio Refinement (BETA)
implemented these parameters at a certain level, you can try that job too. It will be a bit faster.

You can also download/install the older version, even in cluster environment.

3

Thank you.
I will try these out, or try to get the older version if this doesn’t work out.