Best Practices for Creating Galaxy Tools

This page is written by experienced tool developers and contains information about what practices in Galaxy tool development tend to be the most successful ones.

Why you might read this

If you need to maintain existing tools or develop new tools for Galaxy, this programmer oriented guide, contributed by the community, will detail current collected best practice guidelines for building and maintaining automatically installing reproducible Galaxy tools.


Follows a short summary of the key parts when it comes to Galaxy Tools.

  • ToolShed, The Galaxy AppStore - companion Galaxy web server for tools. E.g. or localhost:9009 for Mercurial based source code management and automated installation of all components described below through any Galaxy admin interface.
  • Galaxy Tool - An application specific, XML defined interface and associated documentation exposing any command line application as a form-driven Galaxy tool - e.g. BWA or bamtools. Ideally, as a shareable tool shed repository, supporting automated Galaxy installation with revision/version specific control of dependency binaries for reproducible analyses.
  • Tool Dependency Package - Tool Shed tools in the Tool Dependency Packages category whose names start with package_ such as package_samtools_1_0_0, automate the installation of a specific version of some command line application software that other tools depend on. Each dependency package may be shared by many tools and are only available to users through tool forms that populate command lines and execute them.
  • Datatypes - Galaxy has a flexible and extensible internal representation for specialised data formats such as fasta sequences, fastq short read data or tabular text persisted in Galaxy histories. Tools and packages can extend Galaxy by installing new datatypes when needed.
  • DataManagers - Large scale scientific analyses often involve local copies of canonical reference data such as reference genomes and application specific index files for annotation and mapping in genomics. In most cases these are rapidly evolving and a constant drain on highly skilled resources to keep up to date manually. Data Managers can be built and shared to automate reference data maintenance by the local Galaxy administrators. Data Manager repositories should start with data_manager_.