[BioC] Install details for root level administrators

Martin Morgan mtmorgan at fhcrc.org
Wed Jul 18 07:48:03 CEST 2012


On 07/17/2012 09:08 PM, Paul Johnson wrote:
> Option 3. Figure out the "old fashioned" R instructions to install
> BioConductor packages without using  your scripts.  For example, am I
> correct that the first step is to do this (with R-2.15.1):
>
> install.packages("BiocInstaller", repos =
> "http://www.bioconductor.org/packages/2.10/bioc",
> lib="/usr/local/share/R/bioconductor")

after library(BiocInstaller), biocinstallRepos() gives a vector of 
repositories. Use this, then dependencies (including from CRAN) will 
also be installed. Update (BiocInstaller) when your R is updated. The 
Bioconductor package dependency graph is dense compared to CRAN's, and 
it is more important for packages to be from the same release and 
up-to-date; you loose this by opting out of biocLite (though manually 
invoking update.packages() on a regular basis would be a substitute).

Instead of this, though, why not create an account bioc-installer that 
has write permission to /usr/local/share/R/bioconductor but no other 
special permissions. Ensure that this directory is first in .libPaths() 
(e.g., setting R_LIBS_USER=/usr/local/share/R/bioconductor as a shell 
variable for bioc-installer; see ?.libPaths). Then use biocLite to 
install / update packages in that directory only, without concerns about 
root access. This would seem to be a sensible strategy for R packages 
generally.

Since both R and Bioconductor are versioned, it would make sense to name 
the common directory, e.g., /usr/local/share/R-2.15.1/Bioc-2.10.

Martin
-- 
Computational Biology / Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N.
PO Box 19024 Seattle, WA 98109

Location: Arnold Building M1 B861
Phone: (206) 667-2793



More information about the Bioconductor mailing list