Is the Validation (FSC) was valid indeed?

Hi, all.
After doing a local refinement, I used the output map and mask to do the validation FSC. The result was a little different. In the job validation FSC, the resolution was often lower.

In my opinion, the mask changed. The local refine used a dynamic mask. While the validation FSC used a dynamic mask based on my specified mask. I am afraid that the refine map and the validation curve did NOT match. I guessed maybe I used the output mask to re-do the local refine. The refine map will also probably change. Anyway, I did not care whether the real space map changed. I wanted to know whether it was reasonable if I used the local refine map and the validation FSC curve together.

Hi @Yuqi,

The last iteration of any refinement will show two FSC plots. Both plots will show unmasked, spherically masked, “loose masked” and “tight masked” FSC curves. The “loose mask” curve uses an auto-generated mask with a falloff of 15 Å.

For the first plot: the “tight mask” curve uses the dynamic or static mask provided to the job. Where you may be seeing a discrepancy is in the second plot: it shows an FSC curve using an auto-tightened mask as the “tight mask”. It is not the same as the dynamic mask – it is usually somewhat tighter. This mask is available as a low-level output of the job (volume.fsc_mask_auto). If you override the mask input to the validation (FSC) job with this mask instead, you should see the plots agree.

Let me know if this doesn’t resolve the discrepancy!


1 Like