Difference between revisions of "Software:Application"

Latest revision as of 13:44, 5 January 2023

OBSOLETE

The Centre for Advance Computing has a large amount of software pre-installed for use on our clusters. Here we give an overview of software installed on the Linux platform (CentOS). Note that software that was installed for the Solaris platform is not documented here, as that platform is being discontinued.

Table of Software on the SW Linux (CentOS) Platform

The following software can be accessed from the swlogin1 (login) node or through the Grid Engine scheduler. Most software may be accessed through usepackage entries which add software and libraries to your $PATH. As an example, "use anaconda3" replaces the default version of Python with the copy of Python provided by Anaconda. Which copy of software you are using may be verified with "which X", where X is the name of the executable. Note that some package entries are purely short cuts provided for convenience, and will add the default version of that software to your $PATH. The entries these short cuts "point to" are indicated in the "Points To" column.

NOTE: Unless explicitly specified here, it is generally best to run software using the usepackage entries (i.e. "use xxxxxxx") instead of attempting to call the software directly from its absolute location in /opt. The reason for this is that "use xxxxxxx" often loads required dependencies.

use java8; 
java -jar /opt/picardtools/2.0.1/picard.jar

Language-specific software package installation

Certain programming languages require the use of multiple add-on packages to reach their full functionality. If we are missing a package you need, there are two options: either request it be installed email us at cac.help@queensu.ca or perform a local install. Both methods will give you a working copy of the software package

Perl

To install Perl modules to a local directory, use the following bash commands to create a localized install of whatever modules you may need. It's actually not as complicated as it looks.

# install local::lib
wget http://search.cpan.org/CPAN/authors/id/H/HA/HAARG/local-lib-2.000018.tar.gz
tar -xzf local-lib-2.000018.tar.gz
cd local-lib-2.000018
perl Makefile.PL --bootstrap
make test && make install

# setting up appropriate environment variables so that perl knows about our new ~/perl5/lib directory
cd ~                
echo 'eval "$(perl -I$HOME/perl5/lib/perl5 -Mlocal::lib)"' >> ~/.bashrc
source ~/.bashrc

# check that local::lib is indeed installing to the right directory, you should see a bunch of paths beginning with ~/perl5/lib/perl5/ get printed out
perl -e 'print "@INC"'

Installing Perl modules from CPAN will now allow you install whatever you need. For an example of using CPAN to install BioPerl, see below. This part requires a bit of baby-sitting, just hit enter whenever a prompt comes up.

cpan
install CJFIELDS/BioPerl-1.6.924.tar.gz
q

Python

We highly recommend using one of the two Anaconda installations on the SW cluster. These already have a large number of pre-installed packages and will serve you well. However, if they are missing something you need, you can make a local install using the following instructions:

use anaconda3 #or "use anaconda2" for python 2.7
pip install --user packageName

R

When using R, make sure to use the centralized install first with "use R". The system R version on the login node has a slightly different set of installed libraries from the versions of R found on the production nodes, which can result in dependency issues. Adding "use R" to your scripts will avoid this.

When installing R packages from CRAN, it's easiest to manually specify the CRAN mirror to download from with:

install.packages("packageName", repos="mirrorURL")

where mirrorURL is one of the repositories listed at [1]. It will tell you at some point that the directory is not writeable and ask you if you want to make a local library instead- select 'yes'. The package should install normally and be ready for use.

When installing packages from Bioconductor:

source("http://bioconductor.org/biocLite.R")
biocLite("packageName")

If you see an error about "URLs are not supported", simply change any "https" URL to "http" and it will work. Again, it will ask you to make a "personal library", select 'yes'.

Using your own software

It's very easy to use software that is not pre-installed on the systems. Simply specify the absolute or relative path to your software in a script and it will run. For example, if you had a program called "test" located in ~/bin/, you could easily invoke it with "~/bin/test" in your scripts.