cnvGWAS migration

[lgeistlinger]

Hi @viniciushdasilva - I think it will be good to have a timeline for the migration. Will you be able to move things over until May 18th? That's the deadline for changes included in the upcoming Bioc3.13 release https://bioconductor.org/developers/release-schedule/? After that it will be another 6 months for the package being in the current state, so I think we should make an effort to meet the deadline.

[viniciushdasilva]

Agreed. I will merge a new cnvGWAS version until May 16th. Thus, we have time for some cosmetic improvements.

[viniciushdasilva]

Hi @lgeistlinger. I just updated a newer package version that performs linear regression without PLINK. Linear mixed models as present in dev branch will require further discussion as it depends on lme4qtl, which is not on CRAN.

[lgeistlinger]

Cool - thanks, Vinicius! I just pushed your changes on to Bioconductor.

Linear mixed models as present in dev branch will require further discussion as it depends on lme4qtl, which is not on CRAN.

I think here your options reduce to (i) convince the lme4qtl authors to submit to CRAN (which I remember they were not reponsive to), or (ii) copy the corresponding part that you need from lme4qtl into CNVRanger.

[viniciushdasilva]

We could try to convince the authors again. But they refused last time we did.

[lgeistlinger]

How much copying of code would it involve do you think? I think they are pretty set on not submitting to CRAN / Bioc.

[viniciushdasilva]

dev branch needs core functions from lme4qtl. We could try to invite lme4qtl authors to the next CNVRanger version and ask if we can copy the necessary code from lme4qtl into our package, at least until they manage to upload their package to a proper repository. Nevertheless, we could cite their publication in the cnvGWAS function help page and in our vignette explicitly.