Fail in global ctf ref

CTF Refinement
1

the v 4.5.1 failed in global ctf refinement. (local ctf refinement works) Your help will be appreciated! :Lan

Traceback (most recent call last):
File “cryosparc_master/cryosparc_compute/run.py”, line 115, in cryosparc_master.cryosparc_compute.run.main
File “cryosparc_master/cryosparc_compute/jobs/ctf_refinement/run.py”, line 163, in cryosparc_master.cryosparc_compute.jobs.ctf_refinement.run.run
File “/home/cryosparc_user/software/cryosparc/cryosparc_worker/cryosparc_compute/exposure_groups.py”, line 58, in check_particles_groups
retval = check_field_in_group(group, egid, fullname)
File “/home/cryosparc_user/software/cryosparc/cryosparc_worker/cryosparc_compute/exposure_groups.py”, line 45, in check_field_in_group
if do_assert: assert False, “Field %s is not constant in group %d with %d items, should be %s” % (fullname, egid, len(group), str(firstval))
AssertionError: Field ctf/cs_mm is not constant in group 8 with 22713 items, should be 2.9550881

2

Hi Lan,

it is indicating that some particles within an exposure group have different Cs. This could happen if you refined Cs outside Csparc, with per-group CTFrefine in relion for example, then imported into CS as a single group.

Cheers
Oli

3

dear Olive, thanks for the comments. this input was directly from a local refinement. Lan

4

Reset Cs (it’s an option in the Global CTF Refinement job) and re-refine to see if it re-occurs.

Also, at what point did you do Cs estimation? What microscope are you using? A Cs of 2.95 is pretty unusual, and if your data is from a Titan that’s a ~10% error (where I would expect the estimation to be wrong rather than the microscope that far off!) although Cs drifting that far can be due to pixel calibration being wrong, so check that also.

1 Like
5

Thank Olver and rbs_sci for your comments. I did Particle Subtraction and tried to re-select more particles from the same imported micrographies. I performed the same treatment for both sets. after combining the two sets of particles, the resolution and orientation distribution are improved. However, when I run global crt refinement, it failed for the combined particle set but works for the two subsets individually. Should I ignore this problem and continue the combined set? Your future insights are appreciated.
Lan

6

If you did global CTF refinement independently on the two datasets, chances are the exposure group IDs are the same (or overlap) but the calculated parameters are different between the two datasets. Hence, when you combine them, CryoSPARC complains about things not being consistent.

7

would you suggest ignoring this issue and continuing the combined data? :slight_smile:

8

If you are at a resolution where you can start to think about fitting models, I would check the pixel calibration as an estimated Cs of 2.95mm makes me nervous.

1 Like