Fix dependency for plant_tribes_kaks_analysis 1.0.3.0

[jennaj]

This tool suite is available at usegalaxy.org, not EU or AU.

Failure results in putatively successful "green" datasets but the output contains a log message instead of actual output. Using options that involve BLAST trigger the issue, default options do not.

See dataset 13 here https://usegalaxy.org/u/jen-galaxyproject/h/test-kaksanalysis-ksdistribution

Full details: mvdbeek/usegalaxy-tests#49

[gregvonkuster]

I've sent a note to the PI whose lab built the underlying PlantTribes tools to see if they are still maintaining / using them or if he would rather deprecate them on main. I'll let you know what I find out.

[gregvonkuster]

I heard back from the PI that a couple of papers are soon to be published for PlantTribes, so it would be great if we could keep these tools maintained for Galaxy. I think the issue here is a configuration glitch on TACC nodes where the job runs, and I see in the slack thread that Nate will be taking a look at it.

[natefoo]

The BioContainer includes a version of ruby that is missing a dependency (libjemalloc.so.1) because a newer version of jemalloc is installed that provides libjemalloc.so.2 instead. I don't know what pulled in and called ruby since it's not a direct dependency of plant_tribes_kaks_analysis. I am looking in to building a new version of the 1.0.3 container with a new version of ruby, but it is worth mentioning that the 1.0.4 container does not appear to have this problem. Is there any chance the wrapper could be updated for 1.0.4, if it's not a significant change?

[natefoo]

PR for the container rebuild is BioContainers/multi-package-containers#2383

[gregvonkuster]

Is there any chance the wrapper could be updated for 1.0.4, if it's not a significant change?
I'm the most likely candidate for doing this. I'm currently under water, but I'll try to get some time to look at this.

[natefoo]

Looks like we've gone both routes. I have a fixed container image for 1.0.3 that should be deployed by tomorrow, and Greg updated the wrapper for 1.0.4, which we can install.

[jennaj]

Hi @natefoo @gregvonkuster

Is this ready to install? Are other tools impacted? I don't see the update in the MTS (still the 2018 version).

A user reported an issue with another tool from this suite: GeneFamilyClassifier. The output directory is red without much other information. Link to the history: https://usegalaxy.org/published/history?id=014cfd0d6266de50

• toolshed.g2.bx.psu.edu/repos/greg/plant_tribes_gene_family_classifier/plant_tribes_gene_family_classifier/1.0.3.0
• Command Line python '/jetstream2/scratch/main/jobs/48801603/tool_files/gene_family_classifier.py' --input '/jetstream2/scratch/main/jobs/48801603/inputs/dataset_84c764ed-883d-407a-a67d-ef67e6de5bad.dat' --scaffold '/cvmfs/data.galaxyproject.org/managed/plant_tribes/scaffolds/26Gv2.0' --method orthofinder --classifier both --config_dir '/cvmfs/data.galaxyproject.org/managed/plant_tribes/scaffolds/26Gv2.0' --num_threads ${GALAXY_SLOTS:-4}
• the output from an earlier run (that probably has better suited inputs) was able to produce a bit more: Tool Standard Error /jetstream2/scratch/main/jobs/46487041/tool_script.sh: line 22: python: command not found

[natefoo]

I've updated the PlantTribes tools to their latest versions.

Yeah, toolshed.g2.bx.psu.edu/repos/greg/plant_tribes_gene_family_classifier/plant_tribes_gene_family_classifier/1.0.3.0 is failing due to a missing dependency on Python. The tool's utils.py will probably need to be added to <required_files> as well for Pulsar.

[gregvonkuster]

Thanks a bunch @natefoo !