HPSCC setup
Table of Contents
1 Introduction
The HPSCC will be a collection of software for the Biostat cluster maintained by a group of non-sysadmins. It includes R, tools for high-throughput sequencing and additional utilities.
2 Announcements and help
Announcements regarding this set of tools is send to the bit-help email list. It is a very good idea to subscribe to this list to be up to date with what is happening. Contact Marvin for subscription.
Questions should preferably be sent to bithelp@jhsph.edu, or perhaps to Kasper Daniel Hansen (khansen@jhsph.edu).
3 What is there
3.1 R
This includes many versions of R as well as R-devel
. R-devel
is built every morning at 6am and
all installed packages are updated. Depending on the R release schedule, other versions of R may be
updated daily at the same time.
The specific setup file is setup.R.sh
. It is included in setup.all.sh
.
3.2 ShortRead Tools
This includes
bowtie
bwa
maq
samtools
tophat
cufflinks
BOOST
(dependency)
and possibly other programs as needed.
If a new version has been released for one of these tools, and we do not have it installed, please contact bithelp@jhsph.edu or khansen@jhsph.edu to get it fixed. It is probably because we did not notice it.
The specific setup file is setup.shortread.sh
. It is included in
setup.all.sh
.
3.3 Utils
This includes
Subversion
Mercurial
XZ utils
The specific setup file is setup.utils.sh
. It is included in
setup.all.sh
.
3.4 Libraries
This includes
Boost
NetCDF
Eigen
The specific setup file is setup.libraries.sh
. It is included in
setup.all.sh
.
3.5 Emacs 24
Including ESS
, Orgmode
and AUCTeX
in their latest versions
The specific setup file is setup.emacs.sh
. It is not included in
setup.all.sh
.
4 How to use it
Inside the directory /hpscc/usr/local/setup/
(on the cluster) are a
number of files of the format setup.*.sh
. Each of these files
contains commands to set up the appropriate suite of programs (such
that setup.R.sh
contains the information to set up R). By setup we
mean specifying the right paths for executing programs, searching
man/info pages, linking etc.
Each of these files simply needs to be sourced in order to work (an example in the next paragraph).
One file is special: setup.all.sh
which makes a "standard" setup
(including all future programs we install). If you are willing to
trust us, all you need to do is to include the following lines of code
inside your .bashrc
file in your home directory:
if [ -f /hpscc/usr/local/setup/setup.all.sh ]; then source /hpscc/usr/local/setup/setup.all.sh fi
5 About the R installation
R
will always point to the latest stable release. There will also
be R-$(VERSION)
for specific versions of R as well as R-devel
.
R-devel
(and its installed packages) is updated daily at 6am east
coast time. In fact, you
might be running R-devel
on the cluster while we update it. This
might crash your R-devel
session (although we have not seen this
happen ever). But consider yourself warned and do not
complain if you lose work.
R
(the stable version) will seldom be updated (probably around every
time a minor version is released). Its packages will also be updated
seldom, although it happens occasionally as bug fixes are released (if
you need a package to be updated, contact us).
R
will no longer run on enigma2
. If you are running R
inside of
Emacs using ESS, and you are unable to start R
(and you are
constantly being prompted for the start directory) you are probably
running on enigma2
.
6 Installing R packages as a user (not administrator)
If you have an R package you want to use, and you believe there is even the smallest chance that someone else is interested, the easiest is to ask to get it installed system-wide. Email khansen@jhsph.edu.
Otherwise, you should be able to use install.packages()
and R CMD INSTALL
out of the box. This
requires that you are NOT setting RLIBS
in your .bash_rc
nor using .libPaths()
in your
.Rproile
.
Please note that installing R packages is version specific, so you need to install it on every version you use.