We need to report the B-factor value used in map sharpening.
cryoSPARC website indicates that the map-sharpening process starts with the (negative) B-factor value calculated in the Guinier plot. Is there a log file that indicates this B-factor value used initially and how it changed in the final iteration of the map sharpening process? I cannot find any log output in the map sharpening tool or during auto-sharpening within the NU or local refinement processes that reflects this negative B-factor value.
Or is there a metadata file associated with the map that reflects this?
at the end of every iteration of refinement, it produces FSC curves, and below that is a guinier plot. the plot has a positive B-factor value. The map is sharpened at the negative of that number, and you can run sharpening with similar or slightly adjusted (up or down) values and see the effect.
I usually estimate that CryoSPARC (by default) is about a factor of two sharpened above what RELION would estimate for a result from the same data (this has been borne out through testing a couple of dozen datasets of varying types and sizes) but this is partly due to how they both do the calculations. I’ll dig the link out later if no one else has linked it in the meantime, it was a discussion on this forum.
But equally, I usually think RELION over-estimates a touch as well. I’ve also seen a significant improvement in CryoSPARC with RBMC particles vs non-RBMC particles - more than I would have expected in RELION, for example.
How good both of their estimates are depends on how far the resolution extends beyond the inflection point (10 Å for RELION, I think it’s the same for CryoSPARC but drawing a blank right now) and the poorer the resolution the more extreme the b-factor estimate. I would not trust estimated b-factor for any map worse than ~5 Å.
The current “best practice” is estimate b-factor with a Rosenthal/Henderson plot, but for very large datasets that is computationally expensive.