How to install a specific version of R package in Bioconductor?

Question

[xieshaojun0621]

I'm using DESeq2 to analyze RNA-seq data. Since my colleague used version 1.8.1 under R-3.2.2, I'd like to use the same version.

But when I tried to install DESeq2 using: source("http://bioconductor.org/biocLite.R") biocLite("DESeq2"), it automatically installed DESeq2_1.10.1.

My questions is how I can install DESeq2_1.8.1 under R-3.2.2.

[Devon Ryan]

You might get lucky and find this available via bioconda. There are a bunch of bioconductor packages of various versions available there (this is mostly for integration with Galaxy, but there's no reason you couldn't use them). The benefit of this is that the dependencies should all be there then.

Answer 1

[tiago211287]

You can download the tar.gz from this link:

http://pinot.fhcrc.org/packages/release/bioc/html/DESeq2.html

and install it manually in your R or Rstudio. In the 'Install Packages' tab, instead of use the Repository(CRAN), choose your file in your computer.

[igor]

Caveat: then you have to repeat for all the dependencies, unfortunately.

Answer 2

[Joe]

If you can find the package available on github, R has an install_github method https://cran.r-project.org/web/packages/githubinstall/vignettes/githubinstall.html

I've used this for installing particular versions of packages in the past.

[meicaljohnjones]

Dead on, thanks! Still disheartened, I would expect that if you're publishing something in a journal and you don't keep the versions of libraries consistent that your scripts are going to be very brittle over time.

[Joe]

Yeah I should probably have said that in the past I usually used it to install a newer development version of a package than was available on CRAN (which is often a couple of versions behind what you might find on the git repo). How much luck one would have finding older packages I don't know...you might find yourself trawling people's random forks of repos to find a version someone hasn't updated in a while.

Answer 3

[meicaljohnjones]

I've also been looking to do this as well, I think it's kind of an important thing if we're trying to make experiments repeatable. Has anything changed in this regard recently?