Manually deslected tilt still used for tomogram reconstruction

[alain scaiola]

Hi all,

I have some tomograms series where the extremes tilts are way too thick and where there was issue with tracking.

Is there an easy way to remove those from the tomogram reconstruction other than modifying the mdoc file before going to imod ?

I tried to deselect manually those tilt in tomostar mode of Warp, unchecking the boxes in the fourier space view, but they are still used to generate the tomograms.

Thank you for your help.

Best wishes,

Alain

[teg...@gmail.com]

Hi Alain,

If you look into the /TiltSeries/UseTilt node of the corresponding XML file, are the tilts marked as 'False'?

Cheers,

Dimitry

[alain scaiola]

Hi Dimitry,

Thanks for the quick answer. Yes, they are labelled as False, but it doesn't change anything. Could it be a bug on my local installation ?

Cheers

Alain

[teg...@gmail.com]

Oh, I see the bug now: The excluded tilts are weighted as 10e-4, not a flat 0. And in full tomo reconstruction, this weighting is undone after the tilts are combined, so the excluded tilts come back to weight = 1. In sub-tomo reconstruction, the weighting isn't undone, so this leads to correct behavior there. Thus, with the current version, you'll have to exclude the tilts either in IMOD, or right after import by editing the .tomostar file before Warp does any processing (and you'll need to restart Warp for it to re-read the updated .tomostar files). If it's only 1 or 2 tilts, you can probably keep them as they won't have a noticeable effect on template matching. Sorry!

Cheers,

Dimitry

[alain scaiola]

OK, got it. Thanks for the quick answer. I'll exclude the tilt in the text files then !

Have a nice week-end

Best,

Alain

[Reika Watanabe]

Hello Alain,

I am trying to remove the lousy tilt images as well. Finally, how did you do? Did you modify mdoc file? 
If some of you have tried other ways, let me know.

Thank you very much!

Reika 

[Arjun Bhatta]

Hi Reika,

I have just been modifying the mdoc files to remove bad tilts. When using AreTomo for alignment, I just use the list in *_DarkImages.txt file to remove dark tilts.

Attached here's a little python script I wrote which might be useful for me if you are using AreTomo. If you use Imod, you would need to somehow generate a txt file with the list of excluded tilts (or modify the script to read included tilts from the taSolution.log file). 

All the best,

Arjun

[Alister Burt]

Hi all,

Just an FYI which might be easier than editing mdocs, you can exclude bad 2D images in Warp and they won’t be included in stacks generated for IMOD/AreTomo 

https://teamtomo.org/walkthroughs/EMPIAR-10164/2d-preprocessing.html#deselect-bad-images

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 on the web visit https://groups.google.com/d/msgid/warp-em/111f8fe1-cad7-4f38-ac7d-c1cbee193953n%40googlegroups.com.
<DelMdocTilts_aretomo.py>

[Reika Watanabe]

Hello Arjun and Alister,

Thank you very much for your excellent input and for sharing your script, Arjun! So generous of you! I have never tested AreTomo, and it seems super popular now. Do people get better subtomogram average data with AreTomo? It would be great if you could share your experience! 

As suggested by Alister, deselecting bad 2D images seems super efficient. I have been reading our favorite teamtomo tutorial, but I naively believe that I need to modify mdoc to adjust the missing projection images. 

I will test and come back if I encounter the issue!

Thanks a lot! 

Reika 

To view this discussion on the web visit https://groups.google.com/d/msgid/warp-em/88C8012D-12F4-457F-AF08-AA760816184C%40gmail.com.

[Arjun Bhatta]

Hi Reika, Hi Alister,

Sorry for talking a while to reply. I agree with Alister that manually deselecting bad tilts is the most hassle-free way to go about removing a few bad tilts. I just find it more convenient to edit mdocs when there are many tiltseries with high tilt images that are too dark or don't align too well (and you have a list of them); just to avoid the clicking time. But then again, one could probably also just enable "use first XX frames by dose" option in Warp/M for tomogram/sub-tomogram reconstructions to avoid contributions from all high-tilt images without having to edit mdocs.

As for AreTomo, in my (frankly, rather limited) experience, fiducial-less alignment seems to work slightly better than IMOD in my hands and I find it easier to set up for batch/on-the-fly alignments. Since I assume you would go down Warp-M pipeline afterwards anyway, my guess would be you wouldn't be limited in resolution by the software used for initial alignment as long as the alignments are reasonably good for initial picking. But please take what I say with a grain of salt.

Best,

Arjun

[Reika Watanabe]

Dear Arjun

Thank you very much for your reply! 

With my data, M, unfortunately, did not improve the resolution. Therefore I need to improve tilt series alignment.

Anyway, one needs to test with their data! 

Thank you very much for motivating me!

With best wishes,

Reika 

To view this discussion on the web visit https://groups.google.com/d/msgid/warp-em/df521d46-ead3-4bab-b02c-1a5da20400b4n%40googlegroups.com.