IRIDA Pipeline Development

This document describes the necessary steps for integrating new pipelines into IRIDA.


Pipelines in IRIDA take as input data managed by IRIDA and run through a collection of tools to produce some meaningful result. Pipelines are implemented as a Galaxy Workflow and executed using an instance of Galaxy that has been setup for IRIDA. Pipelines are versioned and are stored and distributed along with the IRIDA software. Tools used by a pipeline are versioned and are stored and distributed using Galaxy Toolsheds. In particular, the Galaxy Main Toolshed and the IRIDA Toolshed are used to store and distribute tools for a pipeline.


Currently, there are two pipelines integrated into IRIDA:

  1. A pipeline for constructing whole genome phylogenies.
  2. A pipeline for performing de novo assembly and annotation on genomes.

IRIDA provides support for developing and integrating additional pipelines from Galaxy. This process can be divided into two stages: Galaxy Workflow Development and IRIDA Integration. The necessary steps, in brief, are:

  1. Galaxy Workflow Development
    1. Develop a Galaxy Workflow
    2. Upload dependency tools to a Galaxy Toolshed
    3. Export Workflow
  2. IRIDA Integration
    1. Pipeline Data Model
    2. IRIDA Workflow Description
    3. Additional IRIDA Updates
    4. Run IRIDA

Galaxy Workflow Development

Galaxy provides the ability to organize different bioinformatics tools together into a single workflow for producing specific results. These workflows can make use of already existing bioinformatics tools in Galaxy, or can include customized tools which can be distributed using a Galaxy Toolshed.

1. Develop a Galaxy Workflow

Galaxy provides a built-in editor for constructing and modifying workflows.


This editor allows for the definition of input files and file types, tools and parameter settings for the tools, as well as which files will be used as output from the workflow. More information on constructing Galaxy workflows can be found in the Galaxy Workflow Editor documentation.

In order for a workflow to properly be integrated into IRIDA, the input and output to this workflow must be in a specific format.

A. Input Format

IRIDA currently only supports two types of input files: a collection of paired-end sequence reads in FASTQ format, and an optional reference genome in FASTA format.

For the paired-end sequence reads this must be a dataset collection of type list:paired.


For the optional reference genome, if you wish to use a reference genome, the type must be an input dataset, not a dataset collection.


Please also make note of the names given to each input dataset, in this case sequence_reads_paired and reference, as the names will be used to link up data sent from IRIDA to the Galaxy workflow.

B. Output Format

Output datasets within IRIDA can be of any file type and there can be many outputs for each workflow. Each output should have a consistent name which will be used by IRIDA to find and download the appropriate file from Galaxy. This can be accomplished by adding a Rename Dataset action to each output file. In this case, for the tool PhyML the name is phylogeneticTree.tre.


In addition, each output dataset should be marked as a workflow output by selecting the asterix * icon, in this case both the output_tree and the csv files from the PhyML and SNP Matrix tools have been selected as output.

2. Upload dependency tools to a Galaxy Toolshed

If the workflow being developed includes custom tools that do not already exist in Galaxy these tools should be uploaded to a Galaxy ToolShed to allow for distribution of this workflow. This should be done before building and exporting the final workflow *.ga file, since the ids of each tool in the Galaxy workflow include the name of the toolshed. For example, the id for Prokka, which is used for annotation of genomes, is, which includes the name of the toolshed where Prokka can be found

More information on developing a tool for Galaxy can be found in the Galaxy Tool Development documentation.

3. Export Workflow

Once the workflow is written in Galaxy, it can be exported to a file by going to the Workflow menu at the top, finding your particular workflow and selecting Download or Export. This will save the workflow as a *.ga file, which is a JSON-formatted file defining the tools, tool versions, and structure of the workflow.


IRIDA Integration

1. Pipeline Data Model

The Analysis class is the root class for all analyses. In IRIDA, this class will be used to store the output files once an analysis is complete. No modification or extension is needed for this class unless your pipeline is storing special results which need to be stored in the database.

In order to properly integrate the workflow with IRIDA your pipeline may require adding a new AnalysisType. The AnalysisType enum defines the different types of analyses/pipelines available in IRIDA. You will need to add a new constant here for your particular analysis type. For example:


This links the constant AnalysisType.MY_PIPELINE to the string mypipeline.

2. IRIDA Workflow Definition

In order to integrate the Galaxy workflow with IRIDA, two files must be defined: (a) a Workflow Structure and (b) a Workflow Description file. Both these files should be placed in a directory structure defining the name and version of the workflow. For example, for the pipeline version 0.1, the directory structure should look like:

└── 0.1
    └── irida_workflow.xml

A. Workflow Structure

This is the *.ga that that was exported from Galaxy in a previous step. This file must be named

B. Workflow Description

This is an XML file which is used to link up a Galaxy workflow with IRIDA. It defines the particular Analysis Type a workflow belongs to as well as any dependency tools needed to be installed in an instance of Galaxy. For a very simple workflow, this file would look like:

<?xml version="1.0" encoding="UTF-8"?>

    <parameter name="my_parameter" defaultValue="1">
        parameterName="my_parameter" />
    <parameter name="other_parameter" defaultValue="2">
        parameterName="other_parameter" />
    <output name="tree" fileName="phylogeneticTree.tre" />

A few things to note:

  1. <id> defines a unique id for the workflow. This must be a UUID. A quick way to generate a random UUID on linux is the command uuid -v 4.
  2. <analysisType> defines what type of analysis this workflow belongs to. This string should match the string defined for the custom AnalysisType above.
  3. <sequenceReadsPaired> defines the name of the input dataset in Galaxy for the paired-end sequence reads chosen previously. In this case it is sequence_reads_paired.
  4. <reference> defines the name of the input dataset in Galaxy for the reference file. In this case it is reference.
  5. <toolParameter> defines how to map parameters a user selects in IRIDA to those in Galaxy (defined in the file).
  6. <output> defines, for an output file, a data model name in IRIDA and maps it to the name of the file in Galaxy that was chosen previously. In this case it is phylogeneticTree.tre.
  7. <toolRepositories> defines the different Galaxy ToolSheds from which the dependency tools come from, as well as a revision number for the tool.

For more information, please see the IRIDA Workflow Description documentation. This file must be named irida_workflow.xml.

You can auto-generate an IRIDA Workflow Description irida_workflow.xml file from a Galaxy workflow ga file with irida-wf-ga2xml:

java -jar irida-wf-ga2xml-0.1.1-standalone.jar \
  -i \
  -W WORKFLOW_VERSION > irida_workflow.xml

NOTE: You may need to edit the output from irida-wf-ga2xml to ensure that only necessary tool parameters are kept in the irida_workflow.xml file and that the proper tool revision is used for each tool if this information is not embedded in your Galaxy Workflow ga file.

C. Move Workflow Definition

In order for IRIDA to automatically load up the workflow definition files, the entire directory structure for MyPipeline should be moved to src/main/resources/ca/corefacility/bioinformatics/irida/model/enums/workflows/MyPipeline. So, this should look like:

└── 0.1
    └── irida_workflow.xml

3. Additional IRIDA Updates

A few other smaller steps need to be taken before the workflow is properly integrated into IRIDA. These include the following.

A. Adding a default workflow entry

The file src/main/resources/ca/corefacility/bioinformatics/irida/config/ defines the default workflows associated with a particular analysis pipeline type. This is in the format of irida.workflow.default.[analysis_type]=[analysis_id]. Please fill in the [analysis_type] and [analysis_id] entries for your specific workflow and add this line to the file. For example:


In this case, the [analysis_type] is mypipeline, which comes from the <analysisType> XML tag from the workflow description file. The [analysis_id] is 49507566-e10c-41b2-ab6f-0fb5383be997, which comes from the <id> XML tag from the workflow description file.

B. Adding messages for the UI

Some messages need to be defined in order to display the pipeline in the UI. These are stored in the file src/main/resources/i18n/ and include messages for the title and description displayed in the UI as well as messages for each workflow tool parameter (each <parameter name="<parameter_name>" ... /> in irida_workflow.xml).

The format for these messages must be:



# for each parameter defined in irida_workflow.xml


For the above example with MyPipeline, the messages would be similar to:

workflow.mypipeline.title=My Pipeline
workflow.mypipeline.description=Run my custom pipeline.

pipeline.title.MyPipeline=Pipelines - My Pipeline
pipeline.h1.MyPipeline=My Pipeline

pipeline.parameters.modal-title.mypipeline=My Pipeline Parameters
pipeline.parameters.mypipeline.my_parameter=My Parameter Description.
pipeline.parameters.mypipeline.other_parameter=Other Parameter Description.

C. [Optional] Customize the color of your pipeline

Given your AnalysisType, add an entry to src/main/webapp/resources/sass/pages/pipelines-selection.scss where your AnalysisType is the CSS class. For example, with mypipeline, you would add:

.mypipeline {
  background-color: #c8c8c8 !important;
  color: #222 !important;

4. Run IRIDA

Once you’ve made all the above modifications, you can attempt to load up the pipeline in IRIDA with the command:

mvn clean jetty:run

This should launch an instance of IRIDA on http://localhost:8080. If you log in with admin and password1 you should be able to navigate to the pipelines page, which should now display:


If you select some samples and attempt to run this pipeline you should see:


If you attempt to modify the parameters of this pipeline you should see: