Issue with Auto-expansion

[aeruera]

Hi there,

I am using the SBGrid installation of TomoNet v1.04 to pick particles on viral surfaces from a single tomogram produced in WarpTools. I have encountered an issue with auto-expansion. I picked one point for the top and bottom views and two for the side view, and after combining the mod files, I have two points per contour since TomoNet appears to add a point for single-point contours. But when I get to auto-expansion, I use parameters that I assume are acceptable for this dataset (repeating units 120 A, search range 60,60,60, max initial seed 3 since I have provided only three picks, all other parameters default). I tried 7 rounds and 3 rounds but neither successfully ran. I get the initial message stating "Start expanding for N rounds" and then nothing until at some point I receive "Expand failed for C8_TS_001". There are no error logs in the working directory or any other messages to help me diagnose the problem.

Hoping you can help me work out what's wrong. Thanks for your time.
Thanks,
Alice

[logicvay2010]

Hi Alice,

Thanks for try it out!
Looks like you ran the "combine" in the Manual pick, just to confirm did you also ran he "stalkInit"? That's how by default we generate seed particle files (You suppose to see the output in the subfolder ).
Also, other issue would cause that error, for example, the reference pixel size does not match with the tomograms, or the search box size larger than the reference volume size.
Could you try run the expand command using the command line version? (by enable the print cmd checkbox), the real error message might be omitted by the GUI log window.

Best,
Hui

[aeruera]

Hi Hui,

just to confirm, yes I did run StalkInit but not in the usual way! I picked one point in IMOD for the top and bottom views, two points for the side view, hit Combine to combine the mod files and checked the contours in IMOD of the combined mod file to make sure the contours were sensible (3 contours in total). I tried to run StalkInit through TomoNet but it wasn't possible. If I try to run the StalkInit wrapper, TomoNet 1.04 crashes with an error in the GUI log box saying:

“ERROR = StalkInit error for C8_TS_001! The mod file may have strange number per contour”

In the terminal, it prints: [aeruera@em-viz stalkInit_prep]$ /programs/x86_64-linux/peet/1.18.0/Particle/bin/stalkInit: line 12: 1420999 Killed $NOHUP ${PARTICLE_DIR}/bin/mce/stalkInit_mce "$@" 2>&1

So I ran the command outside of TomoNet by using StalkInit C8_TS_001.mod in the terminal window, which gave the following results:

Starting stalkInit_mce ...
PEET Linux Version 1.18.0 1-January-2025 on 21-Nov-2025 15:44:45.
Copyright 2000-2025 The Regents of the University of Colorado.
MATLAB. Copyright 1984-2024 The Mathworks, Inc.

Input model = C8_TS_001.mod
Reference particle/contour = 0
Output head model = head.mod
Output centroid model = centroid.mod
Output tail model = tail.mod
Output initial motive list = C8_TS_001_InitMOTL.csv
Output rotation axes file = C8_TS_001_RotAxes.csv
Total execution time : 1.948661 seconds

I then moved the output files into ../initParams folder:
C8_TS_001_InitMOTL.csv
centroid.mod
C8_TS_001.mrc
C8_TS_001_RotAxes.csv
I then renamed centroid.mod to C8_TS_001.mod to match my tomoname. After this point, TomoNet accepted the outputs.

I presume perhaps the issue with Auto-expansion originates from this workaround?
I'll try run auto-expansion again with cmd ticked and see what error message it prints. Any thoughts or suggestions welcome!
Thanks again,
Alice

[logicvay2010]

I see, so maybe we have to fix the first error "The mod file may have strange number per contour."
This is caused by when you have odd number of points in before stalkInit, please try to check the combined .mod file, there should be a contour either has only 1 point or 3 points. And the cause should come from your *_side.mod files .
Let me know how does it works.

Best,
Hui

[aeruera]

Hi Hui,

me again with a late update.
I don't think there was an issue with the mod file. The mod file appears to be as expected (one point is placed for the top and bottom picks, and two points are placed for the middle pick. TomoNet then adds two points per one-point contour after I combine the mod file). I'm using the SBGrid installation of TomoNet and I've been told by another user of this installation that they also encounter the same issue with StalkInit. That user also uses the same workaround, which is to run StalkInit outside of TomoNet, move the resulting files manually and rename them, and then re-launch TomoNet. Could you please advise if you ever found out what the issue is with the StalkInit wrapper? StalkInit takes the mod file just fine when run outside of TomoNet, so I think this likely indicates that the mod file is fine. Do you have any other suggestions that I can pass along to the SBGrid software team, in case it's an issue on our end of the installation?

Thanks so much,
Alice

[logicvay2010]

Hi Alice,

Would you mind send me some of your .mod files that had this issue? Better with the associated top, bottom, side views .mod file. I can give you a update once review them.

Best,
Hui

[aeruera]

Sure, I'll email them to you.