search

3dem / relion

Image-processing software for cryo-electron microscopy
https://relion.readthedocs.io/en/latest/
GNU General Public License v2.0
456 stars 203 forks source link

RELION tomogram reconstruction fails (or produces weird contrast) when carbon is in images #1169

Open DrJesseHansen opened 3 months ago

DrJesseHansen commented 3 months ago

version RELION 5.0-beta-3-commit-6331fe

for tilt series alignment I used the command below: `which relion_python_tomo_align_tilt_series AreTomo --sample-thickness-nanometers 137 --gpu 0 --gpu 1 --gpu 2 --gpu 3 --tilt-series-star-file ExcludeTiltImages/job018/selected_tilt_series.star --output-directory AlignTiltSeries/job003/ --pipeline_control AlignTiltSeries/job003/`

for tomogram reconstruction I used command is below: mpirun --np 5 relion_tomo_reconstruct_tomogram_mpi --gpu "" --t AlignTiltSeries/job003/aligned_tilt_series.star --o Tomograms/job021/ --do_proj --w 4096 --h 5760 --d 1000 --binned_angpix 11.05 --j 1 --pipeline_control Tomograms/job021/

Overall the AreTomo tomograms (in external folder) look good, however on tilt series with carbon there is much higher likelihood that alignment fails and the output AreTomo tomogram is poorly aligned. This gives rise to a second problem (below).

When reconstructing the tomograms in relion I notice again that the tilt series with carbon have two issues:

1) alignment is often very poor 2) very dark and light regions appear in the reconstruction (see image below)

So the first issue I'd like to report is that AreTomo is not aligning correctly when carbon is present. Is it possible to turn on patch flags to do local alignments? From what I can tell these options are turned off by default.

The second issue which I presume is problematic and should be reported is the white and black regions that get generated in these tomograms with carbon. This seems quite strange and I'm not sure how to get around it. These regions are also in the even/odd halves. I have rarely also noticed these dark/light areas in tomograms without carbon, maybe it had a tiny bit of carbon edge, I cannot be sure.

image

thank you for your hard work

best

Jesse Hansen

EuanPyle commented 3 months ago

If the alignment is poor then this is an issue with AreTomo rather than RELION. Do you have fiducials in your sample? I think I can see some. AreTomo always performs poorly when gold beads are present in my experimence. I wouldn't read too much into artefacts on the poorly aligned tomograms.