Silent errors with bad install of libglpk

Microbial community profiling HUMAnN
1

I’m working with Humann 3.6. In a batch of about 130 samples, I noticed one kept failing. Looking at the logs, there was the following error:

/usr/local/bin/glpsol: error while loading shared libraries: libglpk.so.36: cannot open shared object file: No such file or directory
Error when running glpsol from MinPath.

Error message returned from glpsol :
load_mps: reading LP data from `/lila/data/brinkvd/isabl/data_lake/analyses/23/53/2353/humann/10194_1717B_H7WG3BBXY_humann3_humann_temp_y09mgf7e/tmpholvcwfa/tmpl2ju_0mi'...
load_mps: name `PATH'
load_mps: 1 rows
load_mps: 0 columns
load_mps: 0 non-zeros
load_mps: 0 right-hand side vector(s)
load_mps: 0 range vector(s)
load_mps: 0 bound vector(s)
load_mps: 7 cards were read
lpx_add_cols: ncs = 0; invalid number of columns



CRITICAL ERROR: Unable to process all thread commands.

Error message returned from command for thread task 44: /usr/local/bin/python /usr/local/lib/python3.8/site-packages/humann/quantify/MinPath12hmp.py -any /lila/data/brinkvd/isabl/data_lake/analyses/23/53/2353/humann/10194_1717B_H7WG3BBXY_humann3_humann_temp_y09mgf7e/tmpholvcwfa/tmpjpeykneo -map /lila/data/brinkvd/isabl/data_lake/analyses/23/53/2353/humann/10194_1717B_H7WG3BBXY_humann3_humann_temp_y09mgf7e/tmpholvcwfa/tmp9cbvr0cm -report /lila/data/brinkvd/isabl/data_lake/analyses/23/53/2353/humann/10194_1717B_H7WG3BBXY_humann3_humann_temp_y09mgf7e/tmpholvcwfa/tmp_ib47cnc -details /lila/data/brinkvd/isabl/data_lake/analyses/23/53/2353/humann/10194_1717B_H7WG3BBXY_humann3_humann_temp_y09mgf7e/tmpholvcwfa/tmp7arp5s9x -mps /lila/data/brinkvd/isabl/data_lake/analyses/23/53/2353/humann/10194_1717B_H7WG3BBXY_humann3_humann_temp_y09mgf7e/tmpholvcwfa/tmpl2ju_0mi


/usr/local/bin/glpsol: error while loading shared libraries: libglpk.so.36: cannot open shared object file: No such file or directory

I thought this was strange as this only occurred in one sample out of the batch, and all were run with the same docker image. So I looked through the logs of the successful runs and saw the same error (it didn’t cause the run to fail), but without the CRITICAL ERROR: Unable to process all thread commands. or the Error when running glpsol from MinPath. parts, just:

/usr/local/bin/glpsol: error while loading shared libraries: libglpk.so.36: cannot open shared object file: No such file or directory

For the root of the issue, I had to run the following to get it to work with the libglpk installed in setup.py.

pip3 install -vvv https://github.com/biobakery/humann/archive/refs/tags/3.6.tar.gz --no-binary :all: && ldconfig /usr/local/lib

In what cases does this result in failure versus a silent failure? What are the implications for the resulting pathways files when the error was silent?

2

Hello, Thank you for the detailed post. HUMAnN should fail with the error "CRITICAL ERROR: Unable to process all thread commands." if any of its threads fail. If you are running HUMAnN through the bioBakery workflows you should see each HUMAnN process fail in the log but the standard error will only show the last process that failed unless you are running with the flag --quit-early. Using the quit early flag will cause the workflow to stop after the first task fails instead of continuing to run all tasks until there are no more tasks available to run.

Thank you,
Lauren