Strange results from local ctf refinement in Gctf wrapper [Bug]




When I run per-particle defocus refinement using the Gctf wrapper, the results don’t make sense - the per particle defocus values are clustered around zero, not near the global defocus values for each micrograph. See results from Gctf log files attached.

As input I have provided the micrographs (from Inspect Picks), and particles (from an Extract Micrographs job). Incidentally when particles are provided from Inspect Picks the job fails with an error. The particles were binned 4x, but I don’t expect that to make a difference, as all Gctf should be using is the coordinates. Is this a bug, or is there something I’m doing wrong?

EDIT: Definitely a bug related to using binned particles I think, as it works correctly with unbinned particles for same dataset. Incidentally, I think the --refine_input_ctf option is broken - the slider is automatically switched on, where it should not be, and if I switch it off it switches itself back on when I submit the job.


GCTF: Bug with per-particle ctf estimation

Hey @olibclarke,

We’ll take a look at --refine_input_ctf and the binned particles issue. Thanks for bringing this up!