r-probmetab getting stuck on "Solving environment:"

[Tomnl]

Hi all,

When trying to install r-probmetab via linux using the following command

conda install -c bioconda r-probmetab

The install gets stuck on "Solving environment:"

However if I specify the build it seems to install with no problem

conda install -c bioconda r-probmetab=1.1=r341hf484d3e_3

Does anybody have any ideas on how to fix this?

FYI:
@RJMW

[PertuyF]

There is currently a performance issue with conda solving an environment. This is due to the large number of packages, versions and builds available in channels defaults, bioconda and conda-forge, which offer a huge number of combinations to solve dependencies constraints.

You can have a look to Bioconda's recommendations regarding solving slowdown in issue #13774.

For instance, in your case using metachannels lowers solving significantly with the following command:

conda create -n test --override-channels --no-pin -c https://metachannel.conda-forge.org/defaults,bioconda,conda-forge/r-probmetab,--max-build-no r-probmetab

But when looking closer, it appears that r-probmetab is built against r-base version 3.4.1 which is pretty old and most certainly makes things harder...
Looking to rebuild without pinnings is this PR #15647

---

EDIT: Stuck with pinning to r-base 3.4.1 due to bioconductor-rcytoscape, best option is using metachannels.

[Tomnl]

thanks @PertuyF,

I think using the https://metachannel.conda-forge.org/defaults,bioconda,conda-forge/r-probmetab option might be problematic when using Galaxy. @RJMW what do you think?

[RJMW]

@Tomnl Yes, this might be problematic for galaxy.

@bgruening any idea how to address this in Galaxy?
@lecorguille did you experience any issues with r-probmetab as part of W4M?

[Tomnl]

After doing some testing locally with conda using Planemo and the Galaxy wrapper https://github.com/workflow4metabolomics/probmetab

Everything seems to pass (at least locally on my system) if we update the requirements to specify a compatible build for r-probmetab

    <requirements>
        <requirement type="package" version="0.4_1">r-snow</requirement>
        <requirement type="package" version="1.0=r3.3.2_0">r-probmetab</requirement>
        <requirement type="package" version="1.1_4">r-batch</requirement>
    </requirements>

I noticed @lecorguille was updating the dependencies for the galaxy tool a few weeks ago in a separate branch and it was failing on travis . So would be good to get his thoughts on this.

[bgruening]

It seems RCytoscape was dropped from BioConductor? This is why we do not have recent builds of r-probmetab as it seems.

Try adding an explicit dependency to r-base=3.4.1

[Tomnl]

Thanks @bgruening,

Seems OK now if I run the following

conda create -n test3052019 -c bioconda r-base=3.4.1 r-probmetab=1.1

Also, I did try to update the recipe to explicitly only use r-base=3.4.1 but I guess you have to explicitly do this at the command line as well when installing

And for the Galaxy wrapper - added r-base 3.3.2 due to other dependencies versions requiring an older R version.

   <requirements>
        <requirement type="package" version="3.3.2">r-base</requirement>
        <requirement type="package" version="0.4_1">r-snow</requirement>
        <requirement type="package" version="1.0">r-probmetab</requirement>
        <requirement type="package" version="1.1_4">r-batch</requirement>
    </requirements>

Regarding the Galaxy wrapper - is there any problem with specifying the builds? To me it seems it is more reproducible to do so, but I am probably missing some important consideration.

[bgruening]

@Tomnl this is ok. Looks good to me.
For more information see here: #13774