error: DLKcat Docker invalid reference format error

[edkerk]

Discussed in #350

^{Originally posted by wshao1 August 23, 2023}
I am trying to create a full ecModel based on Human-GEM 1.16.0, using a modified form of tutorials/full_ecModel/protocol.m. Step 7 (DLKcat prediction through machine learning) breaks at the runDLKcat() command, giving the following traceback:

I've inspected runDLKcat(), specifically the docker run command at line 38, but it's not clear to me why the reference format is incorrect. Perhaps the filepath is specific to whoever created the tutorial, and we need to change it for our own machine?

[edkerk]

@wshao1, could you please provide:

• Your OS + version.
• If Windows, do you have WSL installed?
• Your MATLAB version.
• Your Docker version.

The problem is probably not a filepath that is user-specific. If you've reached Step 7 of the tutorial, then params.path in the model adapter is correct, and that's where the filepath is built from.

Can you also please:

1. Run dbstop in runDLKcat at 38 in the MATLAB command window.
2. Run the runDLKcat() command as you've done previously.
3. Once MATLAB pauses, run ['docker run --rm -v ' fullfile(params.path,'/data') ':/data ghcr.io/sysbiochalmers/dlkcat-gecko:0.1 /bin/bash -c "python DLKcat.py /data/DLKcat.tsv /data/DLKcatOutput.tsv"'] to see what is the command that should be run at that point.

To again remove the dbstop you can run dbclear all in runDLKcat.

[mihai-sysbio]

Thanks for reporting this @wshao1!

I've inspected runDLKcat(), specifically the docker run command at line 38, but it's not clear to me why the reference format is incorrect.

In addition to what @edkerk mentioned above, could I ask you to copy-paste the output of the docker run command here please?

[edkerk]

@mihai-sysbio, when running a terminal command in MATLAB via system, the standard output is shown in the MATLAB command window. So the output from docker run is docker: invalid reference format..

[wshao1]

@edkerk I've tried your 3 steps, but I still see the same error. I assume in step 3 you meant to run system(['docker run --rm -v ' fullfile(params.path,'/data') ':/data ghcr.io/sysbiochalmers/dlkcat-gecko:0.1 /bin/bash -c "python DLKcat.py /data/DLKcat.tsv /data/DLKcatOutput.tsv"']). Without system(), the string is just echoed back.

My system info:

• Windows 11
• Yes, I have WSL installed.
• MATLAB R2021a
• Docker v4.22.0

[edkerk]

No, I did not mean to include system(), I wanted to see the exact string that would be parsed to system(), where for instance fullfile(params.path,'/data') is replaced.

[wshao1]

Ah ok, here is the exact string: 'docker run --rm -v C:\Users\wshao2\OneDrive - University of Virginia\papin_lab\projects\sasco_project_2\deliverables\metabolic_models\analysis\ecHCT116\data:/data ghcr.io/sysbiochalmers/dlkcat-gecko:0.1 /bin/bash -c "python DLKcat.py /data/DLKcat.tsv /data/DLKcatOutput.tsv"'.

[edkerk]

Perhaps the space in the path (OneDrive - University of Virginia) is problematic. Can you move the folder to a different location (and change it in the model adapter's param.path)?

[mihai-sysbio]

@mihai-sysbio, when running a terminal command in MATLAB via system, the standard output is shown in the MATLAB command window. So the output from docker run is docker: invalid reference format..

@wshao1 what I had in mind was to run the Docker command as you copied above directly in the (system) terminal (not from Matlab), in which case Docker would give more info about the error.

Now though, having seen the file path, I am quite sure those spaces are problematic. To verify, could you modify line 38 to include quotation marks " around fullfile():

system(['docker run --rm -v "' fullfile(params.path,'/data') '":/data ghcr.io/sysbiochalmers/dlkcat-gecko:0.1 /bin/bash -c "python DLKcat.py /data/DLKcat.tsv /data/DLKcatOutput.tsv"'])

[edkerk]

I was able to replicate this on Windows, now solved in #352.

[wshao1]

The quotation marks solved it, thank you so much!