Warp to Etomo Tilt Axis Rotation Angle Off

[Kyle Messina]

Hello,

I'm working on a new dataset that just needs some straightforward reconstructions but I'm running into an odd issue. I'm not sure if its related to Warp (probably) or Etomo (less likely).  

I used Warp to generate the .st files for Aretomo but found that the Aretomo reconstructions were really poor so I decided to give Etomo a go.  Using Etomo, I found that the coarse alignments were off at some of the higher tilts (easy correction with Midas) and the patch tracking adequate for the level of detail.  When I attempted fine alignment, I ended up receiving the Warning in the attached image which suggested I use a tilt axis angle of -266.3 rather than -176.3.  This seemed odd since I've always used -176.3 for our scope and I double checked the mdoc file to make sure it was still -176.3 (it was).

I tried swapping out -176.3 for 266.3 just to see and the coarse alignment was great (no Midas needed), the patch tracking was great, and I had no error in fine alignment.  When I got to positioning, I could tell something was off since the X-Y was rotated 90 degrees and no matter how changed the view in Slicer, I couldn't see the X-Z or Y-Z axes.  

So I'm guessing I changed something related to the tilt axis during import or the initial work-up steps (maybe?).  I can always start fresh to see if this fixes the issue, I'm more looking to see what I may have done (or didn't do) unintentionally to avoid doing it in the future.

Thank you,

Kyle

[Alister Burt]

Hi Kyle,

For the tilt axis angle to be correct in the mdoc files SerialEM has to be set up properly - if you're having issues like this you should set the tilt axis angle with --override_axis at ts_import

For this reason, we also allow you to specify --initial_axis when running ts alignment wrappers

Cheers,

Alister

--
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/8a45f2f8-8b35-40eb-bdca-a8a4b9c92f3fn%40googlegroups.com.

[Kyle Messina]

Hello Alister,

If we're collecting the data with Tomo5 and not SerialEM (which is what we are doing at the moment), I assume I should always run ts_import with --override_axis then.  Should I just use the tilt axis angle in the mdoc file or do I need to do some quick math to figure out the correct tilt axis angle?

Thanks!

Kyle

[Alister Burt]

Hi Kyle, 

Knowing the correct tilt axis angle for a given scope should be the responsibility of the user/scope combo - I might remove any auto parsing from mdocs to avoid this coming up again.

You can eyeball the angle from a tilt series but risk being off by 180 degrees which will flip the physical handedness of your tomograms.

Cheers,

Alister

Sent from mobile - apologies for brevity

On Mar 3, 2025, at 07:18, 'Kyle Messina' via Warp <war...@googlegroups.com> wrote:

Hello Alister,

To view this discussion visit https://groups.google.com/d/msgid/warp-em/773aa58c-c836-4c7f-8811-9f8aedac830en%40googlegroups.com.

[Kyle Messina]

Hi Alister,

To make sure I can eyeball/understand the tilt axis angle correctly, is it just the number of degrees counterclockwise from the Y-axis that the tilt axis rotates around (like you described in the previous post)?  So if the tilt axis is running from the lower left of the micrograph to the upper right of the micrograph, then tilt axis angle would be 135 degrees?

Also manually putting in the the mdoc value of -176.26 actually gave me a good Aretomo reconstruction...but its rotated 90 so it looks like I just need to learn the correct tilt axis angle.  Etomo seemed like it was being a bit wonky with coarse alignment but if Aretomo is working, so I can run with that as a check for now.  Thanks again for the help and helping me learn some of minor things I either glossed over or was not taught! 

From,

Kyle

[Kyle Messina]

I forgot to mention that I used the mdoc value of -176.26 using the --override_axis argument. Everything else was identical (to my knowledge).

From,

Kyle

[Ricardo Righetto]

Hi Kyle,

Just a heads up, it might be that Aretomo found by itself a tilt axis angle that is ~90 degrees from -176.26, which would be consistent with your previous report of -266.3. Can you check what is the value of the tilt axis after running Aretomo?

In general the tilt axis angle reported in Tomo5 mdocs is not reliable at all so we just ignore it.

In our case, the Tomo5 mdoc says -174.25 but the correct value for importing EER data into WarpTools is -95.75 degrees.

Best wishes,

--
Ricardo Diogo Righetto

--
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/13530192-154b-410f-9a4d-64455d7001aen%40googlegroups.com.

[Ricardo Righetto]

As for the definition of the tilt axis angle (theta in this figure) in the unaligned TS:

See also:

https://bio3d.colorado.edu/SerialEM/hlp/html/about_formats.htm

It seems to me that the definition of 'Tilt axis angle' and 'RotationAngle' are swapped in Tomo5 mdocs...

--
Ricardo Diogo Righetto

[Kyle Messina]

Hello Ricardo,

Not sure where I can get that value (so feel free to suggest the best way to go about it if this is wrong).  I went to the resulting .st.aln file and I saw a value of + 93.8379 for the Rot column for all tilts.  

And based on your images, our actual tilt axis angle is probably ~100 degrees since our tilt starts just below the x axis on the left and is just above the x-axis on the right. Appreciate the picture! 

From,

Kyle

[Ricardo Righetto]

Hi Kyle,

Yes, it's this +93 value you found in the .aln file. This is the same angle as the -266 you found before, so everything is consistent. Aretomo is finding the correct value for you, and as expected, it is ~90 degrees from what your .mdoc file says.

Now, hopefully the only thing left to figure out is whether you should use +93 or -87 degrees, as the 180 ambiguity changes both the physical handedness and defocus handedness (assuming no other weird axis flips are happening), as Alister mentioned in the other thread. Getting the correct physical handedness is more important, since the defocus direction can be easily assessed and corrected for later.

Good luck!

--
Ricardo Diogo Righetto

[Kyle Messina]

Hi Ricardo,

I set up Aretomo to run before I left work last night and using a value of 93 gave me good results but how do I assess if the physical handedness is correct? Is it something I should be able to eyeball or are there some quick tests I can do (like rotating around X,Y, and Z)? 

Thanks!

Kyle

[Alister Burt]

Hi Kyle,

Warp provides a routine for checking physical handedness by template matching against templates in both hands then comparing peak heights, this requires decent tilt series alignments and will work better the more asymmetric your template is.

Otherwise, subtomogram averaging and comparing the handedness to a known structure

Cheers,

Alister

Sent from mobile - apologies for brevity

On Mar 4, 2025, at 07:38, 'Kyle Messina' via Warp <war...@googlegroups.com> wrote:

Hi Ricardo,

To view this discussion visit https://groups.google.com/d/msgid/warp-em/2363cc52-68b0-4b59-a057-ab0fd2c77701n%40googlegroups.com.

[Kyle Messina]

Hello,

That's right! That step is mentioned in the tutorial. I guess for now I won't worry about that since I just need the reconstructions and no subtomo for this dataset. Thanks again for all the help!

From,

Kyle

[Ricardo Righetto]

Exactly what Alister said, best way is to assess physical handedness by template matching ribosomes and mirrored ribosomes

Good news is that you can do this in a single tomogram from your session/dataset

Alternatively, if you have large chiral structures such as microtubules you can also try to eyeball them

Good luck!

--
Ricardo Diogo Righetto

To view this discussion visit https://groups.google.com/d/msgid/warp-em/73133843-625d-4680-ac16-e52cf8c6b34fn%40googlegroups.com.

[Daniel Cai]

To add onto this discussion, I encountered the same issue as kyle and also fixed it by overriding my tilt axis angle (I believe we were both using Tomo5 mdocs). For me this was going from -175.95 to 85.95.

Strangely, if axis-iter was enabled, my correct tilt axis angle was overwritten and reverted to the incorrect tilt axis angle that was sourced from the mdoc file. (85.95 to -175.95). Not entirely sure on the details of how this option works, but I would suggest anyone facing this issue to turn off that option.

The problem with axis-iter might be caused by warp interacting with aretomo (either through wrapper or through our installation). Running my samples directly in aretomo with axis 85.95 and axis-iter refined my tilt axis to a more reasonable 93. If anyone else has faced this issue, it might indicate the source of this problem.

Thanks,

Daniel

[Tomas Pascoa]

Hi all,

Just to add to this as I missed the original discussion: if everything is configured correctly, to convert the TiltAxisAngle from Tomo5 to SerialEM/IMOD convention, you simply need to do Tomo5TiltAxisAngle * (-1) -90

As far as I understand it, it just has to do with which axis of the image the TiltAxisAngle is relative to, and whether it is measured clockwise or counterclockwise. For example, in SerialEM/IMOD, the TiltAxisAngle is determined counterclockwise relatively to the y axis of the image (as in the figure that Ricardo posted above). So, for data acquired with Tomo5, you really do need to --override_axis using the value calculated with the formula above. I believe that works for all cases mentioned in this thread.

As for physical handedness, as Ricardo and Alister mentioned, you would have to determine it yourself and can invert it (if needed) by rotating the tilt axis angle by 180deg.

Lastly, regarding the issue Daniel mentioned with --axis-iter in the aretomo wrapper, using:

WarpTools ts_aretomo \

--settings warp_tiltseries.settings \

--angpix 7.92 \

--alignz 2000 \

--axis_iter 2 \

--axis -95.75 \

--perdevice 1




the correct refined tilt axis angle was written to the xml files (AxisAngle). Perhaps that worked because we again override the tilt axis angle using the --axis argument?

However, the .tomostar files were not updated with the refined tilt axis angle. At the moment it is unclear to us (I was working together with Davide Tamborrini in Ben Engel's lab) whether the value in the .tomostar files is used at all downstream? As we were not sure, to be on the safe side, we have been updating it with the refined values written to the xml files. @Alister could you please clarify if the tilt axis specified in the .tomostars is used at all downstream, or if the only value that matters is the value stored in the xmls?

Cheers,

Tomas

To view this discussion visit https://groups.google.com/d/msgid/warp-em/f883cc65-689e-4d93-a756-acfde932a149n%40googlegroups.com.