Add gapseq to bioconda

[cmkobel]

Dear Gapseq developers

I made a Bioconda recipe for Gapseq.
https://anaconda.org/bioconda/gapseq

I have a large interest in applying Gapseq to my research. And especially for designing larger pipelines as well as for teaching purposes it is very handy to have an easily installable option of this great tool.

I had some issues installing the Sybilsbml package for R. I'm not entirely sure if it is strictly necessary for basic use cases. And I guess a reasonable workaround for now could be that a user manually installs this package (as outlined in the current conda installation instructions already). The essence is that it is possible to add this package through the bioconda recipe, but I think it would be much better if the original Cran package was fixed instead, as it makes everything much easier for everyone. We can also think about making a script available in the Bioconda package that automatically installs the Sybilsbml Cran-archived package.

As all dependencies have to sit inside conda-forge or bioconda itself, I had to take another detour and make a recipe for r-chnosz as well (https://github.com/conda-forge/r-chnosz-feedstock).

Publishing packages from Github on bioconda uses the "autobump" feature which means that everytime you publish a new release on your Github page, the new version is automatically pushed to Bioconda. I incorporated the gapseq test in the recipe which ensures that only the latest working version is available on Bioconda. You can check out how I made the recipe in more detail here https://github.com/bioconda/bioconda-recipes/pull/47417/files

Anyway - please give the package a spin. I added a basic installation instruction in the README.md so you can try it out - Let me know what you think.

This pull request is in response to issue #60

Best, Carl

[cmkobel]

I should probably mention that this package recipe is only available for linux. If macos support is necessary, we can work on it.

[jotech]

Dear @cmkobel

Many thanks for your kind feedback and for pushing the gapseq conda package!

In the past, we considered creating a conda package but hesitated precisely because of the unresolved sybilSBML issue. Installing sbml, in general, often causes problems related to the complexity of the libSBML standard. The package sybil itself was also removed from CRAN due to optimizer issues, but here, luckily, a conda package exists.
I would argue that SBML support would be central because it guarantees the exchange of models created by gapseq with other cobra environments like cobrapy. Otherwise, users are limited to using R only.

Fortunately, @Waschina created a replacement for sybil called cobrar, which should be ready soon and aims to replace the sybilSBML dependency!

Let me know what you think and thank you again for your efforts to facilitate the installation of gapseq!

[Waschina]

Hi @cmkobel ,

It is great to see gapseq on conda! Thank you!

As @jotech already said, we are currently trying to get rid of all sybil-related code and dependencies. The new cobrar R-package will be used as an interface to libSBML for SBML-export. The package already works well on Linux and macOS, and for gapseq, we won't need Windows support.

We plan to have the first cobrar release via Git Hub in 1-2 weeks. Then, cobrar can already be used in gapseq for SBML export. I have not yet contributed a recipe to conda, but I guess the next steps would be

1. Make a cobrar release
2. Contribute a bioconda recipe for the cobrar package
3. Replace all sybil-dependencies in gapseq with cobrar
4. Make a gapseq release
5. Update the gapseq conda package recipe

Any help in the process is highly appreciated :)

[cmkobel]

Sounds like a good plan. I'll look out for your next release and update the bioconda recipe then. I assume you will update the installation instructions so I know which dependencies to copy over.

[cmkobel]

Do you have the Cobrar release planned? Otherwise I think I would like to use your master version (not release) in the conda package as there has been significant updates to the master branch since last release (https://github.com/jotech/gapseq/releases/tag/v1.2) ?

Best, Carl

[ajakbejdwt]

thank you!

[Waschina]

Hi @cmkobel

A quick update: We are nearly there with the first cobrar release. We are currently replacing sybil with cobrar in the gapseq code on a separate branch and running several tests. We realized that a few minor updates were required in cobrar itself.

I will post here as soon as we have the first stable cobrar release and a bioconda package for it.

[Waschina]

The first stable cobrar version is released: https://github.com/Waschina/cobrar/releases/tag/v0.1.0

A bioconda recipe for cobrar has a pull request, which awaits a review before a potential merge: bioconda/bioconda-recipes#50354

Meanwhile, I am working on final changes to the gapseq cobrar branch for a new release, that has no references anymore to sybil, sybilSBML, glpkAPI or cplexAPI.

Side note: The upcoming version also does not have doParallel, foreach, and CHNOSZ as dependencies.

[Waschina]

"cobrar" is now on bioconda. However, I will need to release an update, because I noticed that infeasible LPs were not well-handled in v0.1.0.

[cmkobel]

OK I'm a bit busy wrapping up my phd thesis, but I'll update the bioconda recipe asap.

[Waschina]

I'd suggest to wait until I updated cobrar to v0.1.1. I'm running final tests.