Local filtering and sharpening after 3D flex

Dear Cryosparc community,
Many thanks for all the useful tools in cryosparc and valuable discussions posted here. It is quite helpful for beginners like me.
I wonder what B-factor should one use to sharpen the 3D flex map ? Can one use B-factor estimated from nonuniform refine ?
Also a local B-factor filtering feature will be helpful for map interpretation of flexible protein if you can include this feature.
Best regards.

Hi @geyang928!

Thanks for the kind words. The Sharpening Tools job in v4.4 now has the (default) option to estimate a sharpening B-factor from half-maps.

You can use this job to sharpen the ‘Flex volume’ output of a 3D Flex Reconstruction job. Make sure to also turn on ‘Generate new FSC mask’ as the reconstruction job doesn’t include an FSC mask in its volume output.


Dear Valentin,
Many thanks for your info.

I also have another question regarding the refinement of high order CTF for 3dFlex jobs.
The reason I ask this is due to the lower resolution of my 3dFlex as compared with NUF which happened in the previous cryosparc version. With the new v4.4, I’ve noticed that 3dFlex will refine high order CTF.

Could you please kindly tell me at which step CTF is refined with the current version. Is it correct just simply cloning my old 3dFlex reconstruction job and shooting a new run without starting from the beginning (3dFlex Data Prep) again.
Thanks a lot for your great effort with the cryosparc community.

Best Regards,

Hey @geyang928!

Following up on this: no part of the 3D Flex pipeline will refine CTF parameters, but as of v4.4, the 3D Flex Reconstruction job can use higher order CTF terms if they are present in the upstream particles.

If you have a pre-v4.4 training job that you’d like to reconstruct with a higher-order CTF fit from NU refine, you can manually connect the particles.ctf output from NU-Refine to the Prepared particles input of 3D Flex Reconstruction (see screenshot below, where J10 represents your NU refinement job).

Hope that helps!

1 Like