Galaxy

Introduction

Galaxy is an open source, web-based platform for data-intensive biomedical research. It aims to make computational biology accessible to research scientists that do not have computer programming or systems administration experience. Although it was initially developed for genomics research, it is largely domain-agnostic and is now used as a general workflow management system in bioinformatics.

The list of tutorials here suggests the range of applications of Galaxy.

Galaxy on Cedar

On cedar we provide one Galaxy instance for every research group. Galaxy installation requires a special setup that needs to be done by Compute Canada (CC) staff. If you need Galaxy for your group please write an email to support team.

Galaxy directory structure

Galaxy is usually installed on the project directory of the group and it contains several sub-directories. The name of the Galaxy top directory is determined by taking the first two character of PI username + "glxy". For example if PI username is "davidc" the Galaxy top directory will "daglxy" and it is located in /project/group name/ were group name is the default group name of PI, eg., def-davidc. Galaxy main directory contains the following sub-directories which is slightly different than the original Galaxy package.

• config: It contains all required configuration files to set up and optimize the Galaxy server. Below we explain some basic concepts of some of configuration files that need to be set up in order to be compatible with our HPC environment, however, we will not cover all concepts.
• galaxy: It contains the core Galaxy package which is written mostly in Python.
• logs: Contains two files, galaxy.log and server.log. All messages during startup or shutdown of the server are written in server.log while all messages during the run are written in galaxy.log.
• plugins: Contains all plugins. In original Galaxy package this directory is located in the galaxy directory.
• tmp: Contains all temporary files that Galaxy needs for compiling and installing tool sheds.
• venv: It is a Python virtual environment directory and it contains all Python package dependencies.
• tool-data: It contains data used by tools. See the samples in data-integration
• tool-dependencies: It contains all packages needed for tool sheds. By default packages in this directory are installed using Anaconda.
• database: It contains input, output, and error files of all jobs that run on cluster nodes.

Galaxy files ownership and modification

All files of your Galaxy instance belong to a "pseudo-account", a shared account that is generated by an administrator at installation time. A pseudo-account does not belong to an individual person, but belongs to a specific group. Everyone in the group can log in to the pseudo-account using SSH keys. The name of the pseudo-account in this case is the same name as the top Galaxy directory explained above, eg., daglxy. In order to modify any file of your Galaxy instance, e.g. configuration files, you first need to log in to the pseudo-account. Before you can log in you must generate an SSH key pair, store your public key somewhere in your home directory, and let the administrator know about that. The administrator will store your public key in an appropriate place, after which you can log in to your pseudo-account.

Galaxy server management

Starting Galaxy server is the first thing that needs to be done by users. Galaxy server should NOT be run on cedar login node or any compute node. We have a dedicated server called "gateway" and it is used for this purpose. It contains a web server with relevant Cedar filesystems, /project and /home directories mounted on it. Users cannot make a SSH connection to this machine due to security reasons, however we have designed a web site on this machine that allows users to start/stop their own galaxy server. The website also allows users to user Galaxy web interface to communicate with the server. To do that please go to the website https://gateway.cedar.computecanada.ca/ and click on Galaxy link. You will be asked to enter your username and password. Your username and password is the same as your computecanada one. Once you authenticate then you will be automatically redirect to your galaxy server manager website where you can manage your server or use Galaxy web interface.

Galaxy configuration

Files in the config directory are used to configure your Galaxy server. Configuring and optimizing Galaxy is tricky and explaining all the configuration files is beyond the scope of this article. For more information we recommend you to take a look at galaxy website documentation. In this section we explained some basic variables that has been set by admin and we strongly recommend not to change them. Below is the list of these variables:

• In file galaxy.yml (the main configuration file):
  • http: contains your unique port number
  • database_connection is the name of your Galaxy database and your database server.
  • virtualenv is the path to a Python virtual environment in the gateway machine
  • file_path, new_file_path, tool_config_file, shed_tool_config_file, tool_dependency_dir, tool_data_path, visualization_plugins_directory, job_working_directory, cluster_files_directory, template_cache_path, citation_cache_data_dir, citation_cache_lock_dir are appropriate paths for tools, tool sheds and dependencies.

The rest of variables and files in this directory can be changed by user.

Running Tool sheds

There are basically two ways to run your tool sheds, 1. run them locally which in this case means running on "gateway" machine 2. run them using job submission to cedar. Please DO NOT run tool sheds locally because gateway has small memory and it is not efficient for running jobs. Configuring Galaxy such that it submit jobs to cedar is in job_conf.xml. Some basic variables in this file are already set by admin for submitting jobs to Slurm. However, they need to be optimized for various types of tool sheds by users. Please first take a look to get families to these variables and the way they used in this files. For example some tools require more memory or more walltime. It is up to the user to optimized the corresponding variables related to each tools.