Hi all!

I would like to run some bioconductor packages on a really old (but still powerful) MacPro (OS X 10.8.5) with R 3.1.3 (I cannot install newer R versions on this machine). Now I installed an old distribution of ShortRead using biocLite(). But when I run

library(ShortRead)

it tells me that the required BiocParallel could not load. Hence I run

library(BiocParallel)

which produces the warning:

'Error in readRDS(pfile):

cannot read workspace version 3 written by R 3.6.2; need R 3.5.0 or newer.'

So package load for BiocParallel fails. I understand that this is because of the old R version used. However, loading the Biostrings package, for instance, works fine, so I was wondering if there is a trick to make ANY bioconductor package load properly (i.e., circumvent this readRDS error issue).

Thanks a lot, Daniel

I don't think you want to trick an old R into loading new packages. Instead, arrange for old R to have access to old packages.

Do this by creating a directory for the old packages.

mkdir ~/R-3.1.3-library

In R, set the library path so that this is in the first position

.libPaths("~/R-3.1.3-library")
.libPaths()  # confirm set correctly

If possible, set the CRAN repository to an appropriate snapshot date, e.g,. in the Microsoft 'time machine' . https://cran.microsoft.com/snapshot/2015-01-01/bin/windows/contrib/3.1/bibtex_0.3-6.zip Looks like for R 3.1.3 you might have been using Bioc 3.0, and from https://bioconductor.org/about/release-announcements/ it looks like the last date of that release was sometime around October 14, 2014. It would be great to install CRAN packages from that date, but I think the best you can do is a snapshot from 2015-01-01. So set

options(repos = c(CRAN = "https://cran.microsoft.com/snapshot/2015-01-01"))

Use biocLite (if that's appropriate) to install packages -- during the installation, you should see (a) the packages from the correct Bioconductor and CRAN repositories and (b) they are installed into the correct library path.

This is only a 'shot in the dark' so you might need to do some additional work -- it will be useful to do this with a small number of packages (Biobase?) to work out the bugs. I'm not sure, for instance, whether that old version of R supports https on Windows, or conversely whether the Microsoft time machine is listening on http.

But this is a very old release; are you really sure you want to go through these steps?