Incorrect pixel size mitigation strategies

Cryo-EM Data Processing
1

As I am doing depositions I noticed a discrepancy in a dataset. When I should have done an import at 0.63275 and binned x2 in patch motion, I instead did an import at 0.6325. Thus, for one set of volumes I have a pixel size of 1.2655 (correct) and one set at 1.265 (incorrect). If my math is correct there is a 0.03951% difference between the two values. I am examining the gross structural features of proteins in the 600 kDa range, and my NUR maps are around 4-6A with my local refine maps (after sym expansion) around 3.7-4.5A resolution. I see two paths forward.

1. I believe I can simply change the metadata in the .mrc header using ChimeraX or RELION tools. An argument can be made that because the difference is so small, this path is allowed, and I can proceed with deposition.

2. Correct the pre-processing, reassign the particles used for refinements to the new micrographs, and re-run the refinements.

Which option is acceptable or encouraged at this point? If option 2 is chosen, apart from a visual assessment in Coot/ChimeraX and vop subtraction in ChimeraX (to compare old and new maps) is there anything that can be done to determine map differences?

2

for such a small change, path 1 is fine (as there will be minimal effect on CTF parameters or dose estimates).

However it is also always worth double checking that the nominal pixel size is correctly calibrated in the first place - if you have a crystal structure of e.g. one domain of your protein, you can do this by measuring correlation after fitting to the map at different nominal pixel sizes (which can be conveniently carried out automatically using phenix.magref).