R

From wiki
Jump to: navigation, search

Introduction

This is a well known open source statistics packages and there is abundant information on the internet about using it.

Marvin's R version is 3.2.1 which dates from mid 2015. Despite this being a little old now (early 2017), it has a very large number of libraries installed, and these are kept up to date.

Nevertheless, the Bioconductor project has been forging ahead recently issues started to appear relating to the age of the the 3.2.1 version.

Installation notes

It has not been possible to compile a version of R-3.3.x because of a dependency on bzip2 version 1.0.6. The previous version of bzip2 (1.0.5) was valid for R-3.2.x versions, but it must be updated from R-3.3.x versions. Unfortunately, the R configure script is unable to find the alternate versions of bzip2 and so falls back onto the system version of bzip2 which is 1.0.5 and must stay that way for system purposes.

teh follwoing link shows my inquiry about this

The new 3.2.5 version

  • compiled with gcc version 4.8.2 (Using Red Hat's Development Toolkit). Make sure it is not version 4.4.7, as this lacks the crucial C++11 additions. Check for this using
    gcc -v
  • linked to OpenBLAS/0.2.19
  • Ordinary JDK/1.6.0 is used (i.e. not 1.8.0)
  • zlib/1.2.8 and bzip2/1.0.6

Note that the OpenBLAS link is a symbolic links from libRblas.so to the libopenblas.so. This must be done

Installing modules

Important modules and those with large dependencies

  • ape
  • adegenet
  • qgraph
  • Rcpp

Rmpi note

install.packages("Rmpi", configure.args = c("--with-Rmpi-include=/usr/local/Modules/modulefiles/tools/openmpi/2.0.1/include/", "--with-Rmpi-libpath=/usr/local/Modules/modulefiles/tools/openmpi/2.0.1/lib/", "--with-Rmpi-type=OPENMPI"))

Installing Bioconductor

Bioconductor is actually a platform of modules which are avilable normally within R, but for installation, need to be installed in a certain manner

source("https://bioconductor.org/biocLite.R")
biocLite()

Explanation:

* the source is a special repostiory loading statement
* biocLite() will install all the base Bioconductor packages if they haven't been installed yet.

From then on, installing the following modules will in their turn install alot of necessary Bioconductor packages (beyond the base installed by the BiocLite() command.