Request: add to bioconda

[wdecoster]

Hi,

Would it be possible to add a GangSTR recipe to bioconda? That makes installation generally lots easier. I haven't added anything more complicated than python scripts, but here are some guidelines...

Thanks,
Wouter

[mbargull]

Hi @wdecoster, @nmmsv,

it has been a while since this issue was opened. But now, thanks to @christopher-schroeder, GangSTR is available via Bioconda 🎉.

We've added a patch to enable us to link to already packaged dependencies:
https://github.com/bioconda/bioconda-recipes/blob/b3d86283385df3ede3fa2c2f2fd0c9d565ffa420/recipes/gangstr/0001-Unvendor-dependencies-use-pkg-config-instead.patch
It would be very helpful to have this behavior available in your upstream sources (e.g., optionally via some -DVENDOR_DEPS=0 flag or similar).
NB: The patch only has

+pkg_check_modules(hts REQUIRED htslib)
+pkg_check_modules(nlopt REQUIRED nlopt)
+pkg_check_modules(gsl REQUIRED gsl)

but having

+pkg_check_modules(hts REQUIRED htslib>=MIN_HTS_VERSION)
+pkg_check_modules(nlopt REQUIRED nlopt>=MIN_NLOPT_VERSION)
+pkg_check_modules(gsl REQUIRED gsl>=MIN_GSL_VERSION)

(with MIN_*_VERSION replaced with proper values, of course) would be even better ;).

Cheers,
Marcel

[wdecoster]

Good to hear!

[christopher-schroeder]

Even if I have initialized this and wrote a simple working recipe, the applause should go to @mbargull for his amazing knowledge about patching, make, cmake and conda.

[mbargull]

I'll take the applause for the last 10 % of the work and tip my hat to you ;).

[gymreklab]

Thanks for adding this!

[nmmsv]

Hi @mbargull and @christopher-schroeder
Thanks for adding the rule to bioconda! I tried installing using conda and ran into a couple of issues. Since I'm not familiar with Conda (and only recently started usingCmake), I decided to ask them here to see if you have an insight.

1. I tried installing Gangstr using conda in an interactive ubuntu 20.04 docker and got the following error. Am I doing something wrong?

$ conda install -c bioconda gangstr
Collecting package metadata (current_repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.
Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.
Solving environment: | 
Found conflicts! Looking for incompatible packages.
This can take several minutes.  Press CTRL-C to abort.
failed                                                                                                                                                                                                     

UnsatisfiableError: The following specifications were found to be incompatible with each other:

Output in format: Requested package -> Available versions

Package libstdcxx-ng conflicts for:
python=3.8 -> libffi[version='>=3.2.1,<3.3a0'] -> libstdcxx-ng[version='>=7.2.0']
python=3.8 -> libstdcxx-ng[version='>=7.3.0']

Package libgcc-ng conflicts for:
python=3.8 -> zlib[version='>=1.2.11,<1.3.0a0'] -> libgcc-ng[version='>=7.2.0']
python=3.8 -> libgcc-ng[version='>=7.3.0']

Package openssl conflicts for:
python=3.8 -> openssl[version='>=1.1.1d,<1.1.2a|>=1.1.1e,<1.1.2a|>=1.1.1g,<1.1.2a|>=1.1.1j,<1.1.2a']
gangstr -> htslib[version='>=1.10.2,<1.11.0a0'] -> openssl[version='>=1.1.1a,<1.1.2a|>=1.1.1g,<1.1.2a']

2. Is there a benefit to using pkgconfig as opposed to building from source like I do in the CMakeLists? I tried using pkgconfig using the changes that you made to the rule, and wasn't able to get it working on my end.

Thanks again! I appreciate your contribution! :)

[christopher-schroeder]

Hey @nmmsv,

the reason this happens is very likely that you did not setup the bioconda channel as stated in the docu and therefore missing conda-forge. You have two options:

1. Setup the default channels by:

conda config --add channels defaults
conda config --add channels bioconda
conda config --add channels conda-forge

This will add bioconda and conda-forge to the default list of channels. Then you can simply use:

conda install gangstr

to install gangstr. This is the prefered way for bioinformaticians, because all the important tools are in bioconda and you dont want to add bioconda and conda-forge to the install parameters every time.

2. Simply add conda-forge to the channels as parameter:

conda install -c bioconda -c conda-forge gangstr

---

In both cases I strongly recommend using mamba instead of conda. You can install mamba by

conda install -c conda-forge mamba

and then use it like you would use conda:

mamba install gangstr

or

mamba install -c bioconda -c conda-forge gangstr

[wdecoster]

Also worth mentioning that it is usually a good idea not to install stuff in your base environment but create a new environment instead
mamba create -n ganstr_env gangstr

[nmmsv]

Oops that makes sense. It works now.
Thank you all for your help in setting this up! I have added installation with conda to the readme page.