Option to output unmasked map in volume alignment tools? (feature request)

Utilities
1

Hi,

When using volume alignment tools, the output map is always masked by the input mask.

This is not always desirable - for example, if one is recentering on one mask, but using a somewhat expanded/altered mask for subsequent local refinement.

It would therefore be convenient if there was an option to output an unmasked volume in volume alignment tools.

Cheers
Oli

1 Like
2

Thanks @olibclarke, it makes sense to have this be a switch, I recorded it

1 Like
3

Thank this will be a good option to have (a switch to select unmasked map output).

Related to this question what I would like to do after a focused refinement is to isolate a smaller volume centered in the focused region (defined by the mask) together with all the associated particles also reentered and cropped for further processing. Is there a way to do this?

Thanks

H.

4

Hi @hsosa,

Thanks for the input, we agree and will make this modification to Volume Alignment Tools.
Regarding your question – after obtaining a volume from a local refinement, you can:

  1. Use Volume Alignment Tools and input the volume, particles, and mask, and enabling re-centering based on the mask’s center of mass. This will re-center the volume/particles/mask (and apply the mask to the volume currently).
  2. Particles can now be re-centered and cropped around their new centers. This step will write out new actual images. This can be done via either re-extracting them (Extract from Micrographs job) or using the Downsample Particles job. Both jobs have options for re-centering based on the alignment shifts, so turn this parameter on, and also specify a new size to crop/extract at, to reduce real-space extent. Fourier cropping can also be done at this point as well.
  3. To get a volume without the mask applied, take the particles output from step 2 and run them through Homogeneous Reconstruction Only.

Let me know if this answers your question,
Michael

5

Yes, this answer my question. Thank you I will try and I am sure new questions will come.

Regarding the masking when re-centering the volume, would providing the coordinates of the center rather than a mask will recenter the volume without masking it, right?
However, I am a bit confused about how to specify the coordinates in the Volume alignments tool
image
It says 3D coordinates of the new center (A or pix)
If it is A or pix how the program knows whether the input is in A or pix ?
If it is in A, would a suitable input be the atom coordinates read from an atomic model placed in the region of interest (e.g. using Chimera) ?

Thanks again

H.

6

Dear @hsosa,

Regarding the masking when re-centering the volume, would providing the coordinates of the center rather than a mask will recenter the volume without masking it, right?

That’s correct

However, I am a bit confused about how to specify the coordinates in the Volume alignments tool

The job now allows specification of the units by adding a suffix of “A” or “px”, i.e., you can choose which unit to use and specify that by adding the appropriate suffix. The parameter description is:

3D xyz coordinates of new center, relative to the corner of the box. Should be a string of the form x,y,z with a suffix of either A or px to specify the units (Angstroms or pixels, respectively); no suffix will be interpreted as pixels. If provided, volume will be shifted such that its center is at this location, and particle shifts will be modified to preserve their alignment. Leave blank to do no shifting. Shifting will be done before symmetry alignment, if enabled. Enter as a comma separated triplet, e.g. 120, 100, 135 px

Note you can view the parameter descriptions when in building mode by hovering over the parameter’s text box :slight_smile:

If it is in A, would a suitable input be the atom coordinates read from an atomic model placed in the region of interest (e.g. using Chimera) ?

I believe so, as long as the origin in the ChimeraX model is set to the corner of the box. The model also may have to be superimposed onto the input volume to the job, to ensure the spatial extent matches – but this is worth verifying that it works