Strange FSC from tight mask during NU Refinement

Hi All,

I got some strange FSC (the red curve in the top FSC Iteration 003 extended to 4.7Å and did not reach zero at the Nyquist) from the tight mask during NU Refinement New (v3.3.2+220518). However, the final auto-tightened mask generated kind of normal FSC.

Does anyone know what possibly happened for the FSC with tight mask? Thanks a lot!

Hi @donghuachen,

With moderate to lower resolution data, it can happen that the “tight” mask default generation procedure results in masks that are “too tight”, which then results in an artifactual FSC as shown in the first plot. These artifacts happen when the edge of the mask “cuts” into the density. Note that in general, the tight mask default parameters were tuned for moderate to high resolution data. In the second plot, the auto-tightening procedure found a tight mask that was looser, hence avoiding the artifactual FSC.

If you look at the volume output, the mask_fsc result will contain the “tight” mask from the first plot. The mask_fsc_auto output will contain the auto-tightened mask from the second plot. You could download and inspect these two in UCSF Chimera if you would like to confirm that this is the case.

EDIT: To prevent this from happening, you can alter the default tight masking parameters in the builder of the refinement job. In this case, you would want to increase the “Dynamic mask far (A)” parameter from its default of 14 Å to something significantly larger, thus making the mask softer.