[Bioc-devel] Whether to copy unexported BSgenome function into multicrispr

Bhagwat, Aditya Ad|ty@@Bh@gw@t @end|ng |rom mp|-bn@mpg@de
Wed Aug 28 11:18:55 CEST 2019

Dear BioC developers,

BSgenome::getBSgenome('mm10') returns BSgenome.Mmusculus.UCSC.mm10::Mmusculus.

But that function also attaches BiocGenerics, S4Vectors, IRanges and Biostrings, which is unfortunate from a keep-the-namespace-clean perspective.

I could instead use the clean alternative
bsname <- BSgenome:::.getInstalledPkgnameFromProviderVersion('mm10')
utils::getFromNamespace(bsname, bsname)

But that, off course, (BioC)Check does not like, since using unexported functions doesn't fit into the R software development paradigm.

So then instead I could copy paste the latter function into my multicrispr <https://loosolab.pages.gwdg.de/software/multicrispr/index.html> package (which is being readied for BioC) and add its author, Herve Pages, as a co-author of my package (after, off course, informing him). Is this the way such things are normally dealt with? Or how would you guys deal with a situation like this?

A function-level question regarding this topic was posted earlier on BioC support<https://support.bioconductor.org/p/123709/#123711>. The current question is more of a package development question, though, and I thought it would be nice to get feedback from more experienced BioC developers :-)

Thank you for your feedback!


	[[alternative HTML version deleted]]

More information about the Bioc-devel mailing list