fixed compatability with CTF volume reconstructed by WARP

[alncat]

Dear PyTOM developers,
I noticed that there is a small bug when performing local alignment using the subtomogram and CTF reconstructed by WARP. The CTF volume reconstructed by WARP is of size (D//2+1, D, D) if the subtomogram is of size (D, D, D), i..e, the reduced axis is 0. Therefore, we should expand fourier transforms from reduced representation to full representation by specifying the reduced axis as 0. Hence, I made the modification,

return fftshift(fourier_reduced2full(wedge, reduced_axis=0))

Best Regards,
Zhenwei

[alncat]

Hi @sroet, just checking the status of this PR. Do you have any time to review it?

[sroet]

Hey @alncat, Thanks for proposing a fix and the reminder ping.

About the code, it is mentioned in the docstring of the class on line 1548-1553:

class Wedge3dCTF(PyTomClass):
    """
    This is a class to allow pytom running with 3d ctf volumes from warp or relion as the missing wedge.
    Class handles reading of the wedge file from disk. Before being able to use the 3d ctf from warp it
    first needs to be converted to have the z axis as the reduced fourier space dimension.
    """

This class is for both WARP and RELION files, however for WARP files the user needs to make the z-axis the reduced fourier dimension.

I think the code is working as intended and this would break both our internal representation and the reading of RELION files. I will close this PR, but feel free to reopen/discuss if you don't agree with this.