Worse Viewing Direction Distribution After Re-Extraction Without Binning

[katkona]

Hi everyone,

I’m new to the CryoSPARC forum, so apologies if I’m asking about something obvious, but I’ve run into an interesting issue.

I originally processed my dataset with binned particles, and after ab-initio and heterogenous refinement I obtained a reasonably OK Viewing Direction Distribution, so to improve resolution, I went back and re-extracted them without binning and then did a fresh ab-initio into one class.

Surprisingly, the resulting ab-initio volume now shows a much worse Viewing Direction Distribution than the binned workflow.

This is a little bit unexpected, because the particle set, CTFs, and poses should in principle be identical, just at higher pixel size / box size.

Does anyone know a fix to this? Or is something like this normal?

Thanks in advance!

[CryoEM2]

Ab initio won’t do a great job assigning all angles. If you use the resultant one-class as reference for het refine or NU-refine, it should return to normal distribution. Even better, you can use the binned original volumes with the new unbinned particles and skip a step. It fixes reference volume pixel size automatically.