Resolution in FSC

mbs · March 11, 2019, 4:08am

Hi, I am getting “abnormally high” resolution in 2.4.2 version from FSC curves (see attached) Selection_195|690x330. It looks to me like the programs tries to find intersection of FSC with 0.143 threshold going from high resolution end, not from zero. Is that true?
Thanks,
Michael

stephan · March 13, 2019, 8:43pm

Hi @mbs,

Can you reattach your image? It doesn’t seem to be working.

mbs · March 14, 2019, 1:33pm

Continuing the discussion from Resolution in FSC:

Hi @stephan,
sorry for that, i hope this time it would be visible.
Michael

olibclarke · March 15, 2019, 12:54pm

Yes - cryoSPARC calculates res as the last time the FSC crosses 0.143, rather than the first, I’ve reported this as well, it is evidently a bug, hopefully fixed soon

mbs · March 17, 2019, 4:32am

I agree, and that is causing problems with refinement.

ali.h · March 29, 2019, 4:16pm

Hi all,

This issue will be fixed in an upcoming release.

Best,
Ali

mbs · April 3, 2019, 10:45pm

Thanks you Ali,
Could you send me the link to beta version?
all the best,
Michael

ali.h · April 6, 2019, 3:03pm

Hi @mbs, please check your inbox for a direct message from me!

Best,
Ali

spunjani · May 15, 2020, 5:31pm

@olibclarke This is fixed in v2.15

olibclarke · May 24, 2020, 12:47am

Hi @spunjani - it does not seem to be fixed completely unfortunately - see attached from non uniform refinement job in v2.15:

olibclarke · May 25, 2020, 2:59pm

@spunjani is it possible this was only fixed in the “new” refinement, and not in NU refine with legacy codepaths (without higher-order CTF switched on)?