Specifying a wedge cutoff radius in the wedge section of a particle list can cause issues with running FRM alignment. The issue is likely an unlucky combination between the wedge cutoff and the FRM start frequency. It results in the correlation nominator, denominator1, and denominator2 being zero (pytomc/sh_alignment/frm.py in frm_align() at line 1110).
I found a problem for the following parameters:
Wedge cutoff radius set to 30.0, angle=[30.0,30.0]
FRM start frequency set to 8 px, polynomial order=[8,64]
Dataset:
pixel size 2.62 A
subtomogram binning 4x
subtomogram size 50
particle diameter 300 A
Also tried running with unbinned subtomos and same issue occured.
Example data for this issue can be found on angstrom in the following locations
base folder = /data2/mchaillet/processing/mitopaper/tomo04
{base folder}/frm_cutoff and {base folder}/particles_cutoff, provide the avalaible data to reproduce the problem
{base folder}/frm_nocutoff and {base folder}/particles_nocutoff, is where I managed to find a workaround by not specifying the cutoff radius
In the two frm folder are the used job submission scripts (but the pytom paths might need to be changed for testing).
Specifying a wedge cutoff radius in the wedge section of a particle list can cause issues with running FRM alignment. The issue is likely an unlucky combination between the wedge cutoff and the FRM start frequency. It results in the correlation nominator, denominator1, and denominator2 being zero (pytomc/sh_alignment/frm.py in frm_align() at line 1110).
I found a problem for the following parameters:
Dataset:
Also tried running with unbinned subtomos and same issue occured.
pytom 0.994