I'm using R3.5; I see that most of the packages on Bioconductor as with R 3.6.I recently upgraded from R3.4 to 3.5, so it would be difficult to pull myself, scripts, documentation and cluster IT team for R3.6.
I'm interested to install DESeq2. I use BiocInstaller in order to do so.
It prompts me that S4vectors 0.20.1 is loaded but >= 0.21.9 is required by ‘IRanges’
That version of BiocInstaller should tell you that it's deprecated, and that you should use BiocManager. And if you do so, you should do
BiocManager::valid()
Which will tell you what you need to do in order to have a valid installation. And do note that the Bioc version that you are using is most likely 3.8, not 3.5 (they don't match R-versions).
Use BiocManager::valid() to make sure that your installation is correct; likely it tells you that some of your software packages are 'too new', perhaps because you did not install them using BiocManager::install(). Follow the instructions from BiocManager::valid() to bring your installation in line with expectation.
I see; I used BiocManager::valid()$too_new that tells me 9 packages are too new.
I had installed GENESIS "2.14.3" manually which was suggested https://support.bioconductor.org/p/124644/#124651 that led me to install certain libraries compatible with it such as SNPRelate, IRanges, GWASTools, gdsfmt which I installed manually; these are too new. Analyses work fine with GENESIS 2.14 and I'm bit scared to tweak anything in these libraries for the analyses using GENESIS are still ongoing and may wreck tool versions with the analyses.
That is not what was suggested. The suggestion was that you should upgrade R and Bioconductor to take advantage of new functionality. If you go off-script and install things by hand then any support will have to be provided by you - we cannot support the nearly infinite combination of versions that are possible when people mix-n-match like that. It's enough to support the current release version as it is.
If you need to use functionality available only in a more recent version of R / Bioconductor, then you need to install that version of R / Bioconductor. You can see the problems you are causing for yourself by trying to mix different versions.
I see; I used
BiocManager::valid()$too_newthat tells me 9 packages are too new.I had installed GENESIS "2.14.3" manually which was suggested https://support.bioconductor.org/p/124644/#124651 that led me to install certain libraries compatible with it such as
SNPRelate,IRanges,GWASTools,gdsfmtwhich I installed manually; these are too new. Analyses work fine withGENESIS 2.14and I'm bit scared to tweak anything in these libraries for the analyses using GENESIS are still ongoing and may wreck tool versions with the analyses.Thanks.
That is not what was suggested. The suggestion was that you should upgrade R and Bioconductor to take advantage of new functionality. If you go off-script and install things by hand then any support will have to be provided by you - we cannot support the nearly infinite combination of versions that are possible when people mix-n-match like that. It's enough to support the current release version as it is.
If you need to use functionality available only in a more recent version of R / Bioconductor, then you need to install that version of R / Bioconductor. You can see the problems you are causing for yourself by trying to mix different versions.
It is possible to maintain multiple versions of R; some hints are in the BiocManager vignette (at least more recent versions of the vignette ;), e.g., at https://cran.r-project.org/web/packages/BiocManager/vignettes/BiocManager.html)