Tilt handedness

[Huy Bui]

I finally got to use Warp Tool and try to use it for a complete workflow before going to Relion5.

When I imported TS, I didn't use option --dont_invert like Warp Desktop before.

WarpTools ts_import \

--mdocs mdoc \

--frameseries warp_frameseries \

--tilt_exposure 4 \

--min_intensity 0.0 \

--output tomostar

However, after all the steps and after ts_reconstruct, I got a reconstruction with opposite handedness from if I use IMOD to reconstruct or use Relion5 workflow from scratch to reconstruct.

As I know the handedness of my IMOD workflow should be right, how can I deal with this difference from Warp?

Best,

Huy

[Huy Bui]

I just want to add that I tried different settings with the —-dont_invert option, but the reconstruction came out the same. So it is always opposite handedness from IMOD directly or Relion5. Which parameters would lead to a handedness flip of a tilt series?

Best,

H

[Alister Burt]

Hi Huy,

Something about the way you are applying your —dont_invert changes is wrong. Warp has a different convention for what the sign of a tilt angle means internallly - it inverts tilt angles by default when importing tilt series metadata from mdoc files during ts_import to account for this and give a reconstruction which matches IMOD (and by extension, RELION).

Cheers,

Alister

Sent from mobile - apologies for brevity

On Nov 25, 2024, at 21:23, Huy Bui <huy...@gmail.com> wrote:



--
You received this message because you are subscribed to the Google Groups "Warp" group.
To unsubscribe from this group and stop receiving emails from it, send an email to warp-em+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/warp-em/15887aaf-c8dc-4248-accb-c0385c63cfc4n%40googlegroups.com.

[Alister Burt]

tomostar files are the ‘raw data’ used to start processing but the xml file contains the actual value seen by warp if the xml file exists - you are probably updating the tomostar but still taking the same values as before from the XML when reconstructing. You will need to delete the xml, run ts_import then ts_import_alignments to test your angle flipping properly

Sent from mobile - apologies for brevity

On Nov 26, 2024, at 03:54, Huy Bui <huy...@gmail.com> wrote:



Pablo, Can you elaborate a bit?
I can see that if I use --dont_invert my tomostar files have -60 view as 60 degree and without --dont_inver, 60 stay as 60. That makes me very surprised that regardless of what I used, the tomo came out the same handedness.

On Tue, Nov 26, 2024, 1:53 a.m. Pablo Conesa <p.cones...@gmail.com> wrote:

We have fixed this by changing the sing of the angles when calling warptools from scipion

To view this discussion visit https://groups.google.com/d/msgid/warp-em/7F133565-D1F0-436B-BC3A-7E475C8FE8F8%40gmail.com.

[Huy Bui]

I confirmed that delete the xml file fixed it. I though Warp will overwrite after running new but not.

Thanks, Alister.

H

[Pablo Conesa]

We have fixed this by changing the sing of the angles when calling warptools from scipion

El mar., 26 nov. 2024 6:33, Alister Burt <alist...@gmail.com> escribió:

To view this discussion visit https://groups.google.com/d/msgid/warp-em/7F133565-D1F0-436B-BC3A-7E475C8FE8F8%40gmail.com.

[Huy Bui]

Pablo, Can you elaborate a bit?
I can see that if I use --dont_invert my tomostar files have -60 view as 60 degree and without --dont_inver, 60 stay as 60. That makes me very surprised that regardless of what I used, the tomo came out the same handedness.

On Tue, Nov 26, 2024, 1:53 a.m. Pablo Conesa <p.cones...@gmail.com> wrote:

[Philipp Erdmann]

Hey!

In TOMOMAN I put the same.

I multiply all angles with -1 when exporting to warp_tools.

Then everything matches up.

Philipp.

On 27 Nov 2024, at 19:39, Huy Bui <huy...@gmail.com> wrote:



To view this discussion visit https://groups.google.com/d/msgid/warp-em/CADBiWEYm-qPFy%3DbGmJG_uAg4TkxQy2NSyd6Q7dy2LtQNxOCBSQ%40mail.gmail.com.

[B Wimmer]

Hi all,

I've also encountered this issue, see also GitHub Issue #291 (https://github.com/warpem/warp/issues/291).

Basically, previously Warp 1.X did not need the .tlt - file, as far as I can tell, but just read the angles from the mdoc and then flipped them internally. 

Warp 2.X does read the .tlt file, and assumes that the angles are already inverted, independent of the .tomostar file that you create using `WarpTools ts_import`, if that makes sense.

You can see my implementation for the `tomotools` imod2warp / aretomo2warp commands here: https://github.com/tomotools/tomotools/blob/8bd361cc7bd85d60c88603767f7a02c6a36e3e9f/tomotools/utils/sta_util.py#L286C1-L310C11

Hope that helps,

Benedikt

[Huy Bui]

Thanks, Benedikt. I think I did reinvent the wheel here because I did very similar thing that you do by trying to import an already aligned IMOD dataset it.

By the way, since everybody is knowledgeable here, can you teach me about the coordinates in the Template matching produced by Warp 2.0?

I understand it is a fraction of the X, Y, Z dimensions but is there any inversion in certain dimensions vs. other packages or just X/SizeX, Y/SizeY, Z/SizeZ of the coordinate of the particle picking in the tomogram that is produced by Warp?

In addition, the template matching of Warp didn't produce any extra rlnTag such as rlnHelicalTubeID due to it nature. Do you know incorporating these tags will be ok from another picking package to export to Relion5?

Best,

Huy

To view this discussion visit https://groups.google.com/d/msgid/warp-em/5d28437f-2244-42c4-8535-f46197674054n%40googlegroups.com.