Skip to content
This repository has been archived by the owner on Jan 31, 2020. It is now read-only.

How software gets installed in the standalone GMS

Avinash Ramu edited this page Mar 31, 2015 · 21 revisions

##Software is installed into the sGMS in many ways depending on where it comes from and the form it is available in.

###Software that is available publicly as packages

  • Some software is installed directly in the gms/Makefile using 'sudo apt-get install' commands. Some of these are listed directly in the Makefile, while others are listed in gms/setup/packages.lst and gms/setup/install_custom_r/packages.r.lst.

###Software that is used within TGI comes in one of several ways

  • The sGMS itself consists of four git repositories produced by the Genome Institute, gms (the GMS installer), genome (the pipeline code), ur (our custom object relational mapper), tgi-workflow (for constructing workflows to be run on a cluster), and gms-webviews (the sGMS web interface).
  • An archive of pre-compiled binaries specified by $APPS_DUMP_VERSION is installed during the make process. These apps are manually installed software from within the TGI (/gsc/pkg/bio/*) that have not been successfully converted to debian packages yet. These are installed into $(GMS_HOME)/sw/apps which can be accessed by the environment variable $GENOME_SW set in /etc/genome.conf . If you need to modify the packages that are being installed this way, unpack the most recent archive add/remove packages, repackage it with a name containing the new date and move it to the staging directory.
  • An archive of pre-compiled JAVA binaries specified by $JAVA_DUMP_VERSION is installed during the make process. These apps are also manually installed software that have not been packaged yet. They are installed into $(GMS_HOME)/sw/java which can be accessed by the environment variable $GENOME_SW_LEGACY_JAVA set in /etc/genome.conf
  • An archive of debian packages created within TGI are installed globally. This archive is specified in the gms Makefile as $APT_DUMP_VERSION. The creation of this archive of packages happens within TGI and is controlled by the git repo genome-snapshot-deps. Refer here and here for further details. This repo sets up the dependencies for both Ubuntu lucid and precise. From this git repo of dependencies, an apt repo of software is created. The standalone GMS obtains a version of this apt repo by running /gms/setup/stage/Makefile. During installation, the apt repo gets placed here: $(GMS_HOME)/sw/apt-mirror-min-ubuntu-12.04-DATE
  • Some CPAN modules are installed in the Makefile using cpanm
  • Some software is installed in the Makefile by cloning various publicly available git repositories and following the installation procedures of those software (e.g. openlava).
  • A custom installation procedure was created with R to work around problems with the publicly available debian packages for R. We hope to replace that soon. For now the list of installed CRAN and BioConductor libraries installed can be found here in gms/setup/install_custom_r/install_custom_r.sh

The archives mentioned above get placed in the following directory within TGI and from there they are staged to an FTP server that serves them up: /gscmnt/sata102/info/ftp-staging/pub/software/gms/setup/archive-files https://xfer.genome.wustl.edu/gxfer1/project/gms/setup/archive-files

Clone this wiki locally