Error when processing EMPIAR-10500

Import and Export
1

Hi,

When attempting to import movies from EMPIAR-10500 (Falcon 4 data) with the provided gain reference, I get the attached error. Inspecting the header of the gain reference, it is 8192x8192. Is there a reason why EER data is restricted to 4096x4096 for the gain reference?

Also in the EMPIAR entry, it states:

“Please use gainImage_8k_defMasked.mrc as the gain reference without any transformation. Note that the EER gain reference has different convention from those for K2/K3 cameras. That is, the raw values must be divided, not multiplied, by the gain. RELION automatically handles this.”

Does cryosparc also automatically handle this difference in the application of gain references for EER data?

Cheers
Oli

[CPU: 2.72 GB]   Traceback (most recent call last):
  File "cryosparc_master/cryosparc_compute/run.py", line 84, in cryosparc_compute.run.main
  File "/home/exx/cryosparc/cryosparc_master/cryosparc_compute/jobs/imports/run.py", line 823, in run_import_movies_or_micrographs
    gshapey, gshapex = read_gain_reference(gainref_abs_path, shape_only=True, data_format=data_format)
  File "/home/exx/cryosparc/cryosparc_master/cryosparc_compute/jobs/imports/run.py", line 809, in read_gain_reference
    assert gshapex == gshapey == 4096, "Gain reference for EER must be 4096 x 4096"
AssertionError: Gain reference for EER must be 4096 x 4096
2

Hi @olibclarke,

This dataset (10500) is actually collected with a very old EER format specification which TFS changed before actually releasing the F4 camera. The gain reference is no longer 8kx8k for F4, it is now always 4kx4k and also should be multiplied rather than divided (i.e. it is now the same as a normal gain reference from other cameras). So in cryoSPARC we assume that the gain reference and the EER data is of the final production specification. Only this dataset needs to be treated differently - the gain reference must be manually downsampled and you have to take it’s inverse (reciprocal).

Hope that helps!
Ali

3

ah got it, thanks @apunjani!!