Mask used for FSC calculation

[huwjenkins]

I'm trying to understand the difference between --mask_for_fofc and --mask when it comes to the 1/2 map FSC in refined_fsc.json.

Using the command in the ChRmine example the 1/2 map FSC is shifted to higher resolution than the one from RELION PostProcess with the 1/2 maps and mask from EMDB:

However, if I use the command:

servalcat refine_spa --model ../7w9w.pdb --halfmaps ../emd_32377_half_map_1.map\
 ../emd_32377_half_map_2.map --mask_for_fofc ../emd_32377_msk_1.map\ 
--mask ../emd_32377_msk_1.map --pg C3 --resolution 1.95 --cross_validation

I get a curve that much more closely matches the one from RELION.

In the first case which mask is used for FSC calculation?

[huwjenkins]

In the first case which mask is used for FSC calculation?

Answering my own question. As confirmed by playing with --mask_radius it is the mask used in refinement (which is stated clearly in the documentation 🤦)

[keitaroyam]

Yes, exactly. If you want to reproduce the FSC curve from RELION, you can use servalcat fsc --halfmaps half*.mrc --mask mask.mrc which also does the correction through phase randomisation.

--mask option in refine_spa is actually just for a debugging purpose. If you give a soft mask then solvent density in the map will be kept which is not taken into account during the refinement, so you will have smaller map-model FSC (particularly at low resolution, like crystallographic refinement without bulk solvent scaling).