Strange FSC during NU-ref

3D Refinement
1

Hello,

I’m hoping for some advice with refinement strategies. I currently have a really nice map from NUR (my best-looking so far), but with an FSC that correlates poorly in the low res regions.

J78_fsc_iteration_008
J78_fsc_iteration_008506×293 26.3 KB

and has a strange bimodal frequency distribution in the cFSC, showing I am missing info at 3A:
J78_cfscs_half_angle_20_iteration_008_with_autotight_mask
J78_cfscs_half_angle_20_iteration_008_with_autotight_mask558×288 31 KB

However, the same particle stack with homo refinement makes the FSC and cFSC appear normal, but my Map does not look quite as good:

J106_fsc_iteration_006
J106_fsc_iteration_006506×293 27.3 KB

J106_cfscs_half_angle_20_iteration_006_with_autotight_mask
J106_cfscs_half_angle_20_iteration_006_with_autotight_mask558×288 26.2 KB

For reference, I em enforcing C19 symmetry (no relaxation) in both reconstructions. My workflow has consisted of: NUR (with minimize over PPS) of best ab-initio class, followed by local CTF ref, global ctf ref, NUR again with min. PPS, RBMC, NUR, then 3D classification, ending with the reconstructions mentioned above from those 3D classes. Any recommendations for refinement/workflow strategies would be greatly appreciated. Happy to include more information as well.

Thanks!