packages that should be migrated to conda-forge

[bgruening]

A few packages in bioconda, should actually be migrated to conda-forge. This can be a very simple task and is ideally suited to learn both bioconda and conda-forge.

We will keep a list here of packages that we think should be migrated:

• [ ] grep
• [ ] sqlitebrowser
• [ ] perl-* (all perl packages, note that this may prove difficult)
• [X] bucketcache
• [X] represent
• [ ] ant
• [x] argutils
• [x] aria2
• [ ] autolog
• [ ] azure-cli
• [ ] ogdf

Most of these can also be migrated:

• [x] recipes/r-adegenet
• [x] recipes/r-airr
• [x] recipes/r-ampliconduo
• [x] recipes/r-aptreeshape
• [x] recipes/r-asics -> moved to bioconductor
• [x] recipes/r-bc3net
• [x] recipes/r-biom
• [x] recipes/r-blockfest
• [x] recipes/r-breakaway
• [x] recipes/r-canopy
• [x] recipes/r-clvalid
• [x] recipes/r-coenocliner
• [x] recipes/r-corbi
• [x] recipes/r-dbchip
• [x] recipes/r-disco
• [x] recipes/r-ebimetagenomics
• [x] recipes/r-ecodist
• [x] recipes/r-fastqcr
• [x] recipes/r-fateid
• [x] recipes/r-fmsb
• [x] recipes/r-gap
• [x] recipes/r-geiger
• [x] recipes/r-genemodel
• [x] recipes/r-geoaxe
• [x] recipes/r-ggrasp
• [x] recipes/r-goeveg
• [x] recipes/r-gpca
• [x] recipes/r-guilds
• [x] recipes/r-gwasexacthw
• [ ] recipes/r-hardyweinberg
• [x] recipes/r-hierfstat
• [x] recipes/r-imp4p
• [x] recipes/r-ldrtools
• [x] recipes/r-maldiquantforeign
• [x] recipes/r-maldirppa
• [x] recipes/r-metstat
• [ ] recipes/r-micropan
• [x] recipes/r-mmod
• [x] recipes/r-momr
• [x] recipes/r-mvr
• [ ] recipes/r-nabor
• [x] recipes/r-ncbit
• [x] recipes/r-nodiv
• [x] recipes/r-oai
• [x] recipes/r-pegas
• [ ] recipes/r-phyext2
• [ ] recipes/r-phylobase
• [ ] recipes/r-phylomeasures
• [ ] recipes/r-phylosignal
• [ ] recipes/r-phytools
• [x] recipes/r-plsgenomics
• [ ] recipes/r-polysat
• [ ] recipes/r-popgenome
• [ ] recipes/r-popgenreport
• [ ] recipes/r-qpcr
• [x] recipes/r-qqman
• [ ] recipes/r-quorts (duplicate of r-qorts)
• [x] recipes/r-rebird
• [x] recipes/r-rgbif
• [ ] recipes/r-ridigbio
• [ ] recipes/r-rismed
• [ ] recipes/r-rnexml
• [ ] recipes/r-rphast
• [ ] recipes/r-rsnps
• [ ] recipes/r-rtfbs
• [ ] recipes/r-rvertnet
• [ ] recipes/r-sads
• [ ] recipes/r-sbpiper
• [ ] recipes/r-scrm
• [ ] recipes/r-seqminer
• [x] recipes/r-seurat
• [ ] recipes/r-shazam
• [ ] recipes/r-spocc
• [ ] recipes/r-stampp
• [ ] recipes/r-taxonomizr
• [ ] recipes/r-tcr
• [ ] recipes/r-tigger
• [x] recipes/r-traminer
• [ ] recipes/r-treesim
• [ ] recipes/r-vcfr
• [x] recipes/r-wicket

[npavlovikj]

In case this helps, here are few other general packages as well:

• [ ] shapeshifter
• [ ] shapeshifter-cli
• [ ] sharedmem
• [x] r-kinship2
• [x] r-ldrtools
• [x] r-msm
• [x] r-mess
• [x] r-swamp
• [ ] r-syntactic
• [ ] smhasher
• [ ] fitter
• [ ] gromacs
• [X] logbook
• [x] xopen
• [x] libdeflate
• [ ] optbuild
• [ ] fit_nbinom
• [x] pytest-workflow
• [x] cromwell
• [ ] scikits-datasmooth

[simonvh]

I'm moving bucketcache and its deps (logbook and represent) to conda-forge. What is the procedure once they're up and running at conda-forge? Delete them from bioconda-recipes via PR?

[pb-cdunn]

Somewhat related question:

There is a package in Anaconda on a non-standard channel (fragcolor) that I would like to use at build-time for Bioconda packages, but it's not in conda-forge. How can I use it in a Bioconda recipe? (At build-time only.)

Would it be easier if that package were actually in conda-forge?

[dpryan79]

Regarding what to do once packages are migrated, just put them in the build-fail-blacklist file (put the reason why above as a comment). At some point we might due a "grand purge" of such recipes.

[dpryan79]

@pb-cdunn Yeah, try to get it into conda-forge. You can probably download the package from anaconda.org and largely copy what's under info/.

[rhpvorderman]

cromwell

I would say that WDL is more aimed at the life sciences than anything else. But I can understand if this needs to be in conda-forge.

If this is the case then I will migrate pytest-workflow also to conda-forge as soon as cromwell is moved.

[dpryan79]

N.B., recipes can be deleted once migrated to conda-forge.

[marcelotrevisani]

Please, remove pytest-workflow from bioconda

[jenzopr]

I reviewed all r- packages from above and will comment only problematic packages here:

package	comment	suggestion
r-ic10	depends on impute from bioconductor	stays in bioconda
r-swamp	depends on impute from bioconductor	stays in bioconda
r-ontologyplot	depends on Rgraphviz from bioconductor	stays in bioconda
r-stitch	not in CRAN	stays in bioconda
r-sleuth	not in CRAN	stays in bioconda
r-rubic	not in CRAN	stays in bioconda
r-rtassel	not in CRAN	stays in bioconda
r-raceid3	not in CRAN	stays in bioconda
r-quorts	not in CRAN	duplicate - blacklist
r-nanopore	not in CRAN	stays in bioconda
r-lme4qtl	not in CRAN	stays in bioconda
r-gwpcr	not in CRAN	stays in bioconda
r-syntactic	not in CRAN	stays in bioconda
r-metstat	APACHE 2.0 license, but no packaged LICENSE file
r-popgenreport	Lists GPL as license, but no version	upstream issue
r-seqminer	Lists GPL as license, but no version	upstream issue now GPL-3
r-shazam	CC BY-SA 4.0 license, but no packaged LICENSE file
r-tcr	APACHE 2.0 license, but no packaged LICENSE file
r-tigger	CC BY-SA 4.0 license, but no packaged LICENSE file
r-vcfr	Lists GPL as license, but no version	upstream issue now GPL-3

Any suggestion how to continue with them would be helpful. Feel free to edit the comment. Thanks!

[bgruening]

@jenzopr for packages that depend on bioconductor pkgs, we keep them here in bioconda.

For APACHE 2.0 license, but no packaged LICENSE file please drop in a License file in the recipe/folder.

not in CRAN let's copy the meta.yaml from bioconda straight to conda-forge.

Lists GPL as license, but no version not sure what to do here. Assume GPL2? Can we create upstream isues?

[dpryan79]

Actually, for the "not in CRAN" packages, they can stay here :)

[dpryan79]

BTW, @jenzopr thanks for the Herculean effort :)

[pschloss]

It looks like r-breakaway needs to stay in bioconda since it has bioconductor-phyloseq as a dependency. Would there be an easy way to get the recipe back into bioconda?

https://github.com/conda-forge/r-breakaway-feedstock/pull/4#issuecomment-1112406403

[dpryan79]

@pschloss All bioconductor packages will always stay in bioconda. Please ask conda-forge to archive the feedstock.

[calebzulawski]

I need grep for arm platforms, so I opened conda-forge/staged-recipes#22313

[MuellerSeb]

I also think, that all the perl packages should be migrated/available to/on conda-forge. There are for example problems like these: https://github.com/conda-forge/lcov-feedstock/pull/10

That could be sorted out by having the perl packages on conda-forge without creating duplicates as criticized in https://github.com/bioconda/bioconda-recipes/issues/2796

[jvolkening]

I would just like to confirm, before I do much work on this, that it is still the policy to migrate all perl-* packages out of Bioconda, regardless of domain. Thanks.