IRIDA Next Nextflow pipeline specification

This document is in-progress and should be considered as a DRAFT that is subject to change at any time.

This document describes the specification for developing IRIDA Next Nextflow pipelines.

• 1. Starting development
  • 1.1. Developing a pipeline to contribute to nf-core
  • 1.2. Developing a pipeline which won't be contributed to nf-core
    • 1.2.1. Requirements from nf-core
  • 1.3. Requirements for integrating with IRIDA Next
  • 1.4. Other requirements
• 2. Input
  • 2.1. Default samplesheet
  • 2.2. Types of input data
    • 2.2.1. Sample identifier
    • 2.2.2. Input files
      • 2.2.2.1. Example sample input files
    • 2.2.3. Metadata input types
      • 2.2.3.1. Example metadata input types
  • 2.3. Defining the samplesheet structure for a pipeline
    • 2.3.1. Defining sample identifiers in IRIDA Next
    • 2.3.2. Defining selection of input files in IRIDA Next
      • 2.3.2.1. Example samplesheet and JSON schema for an assembly
    • 2.3.3. Defining selection of input metadata in IRIDA Next
      • 2.3.3.1. Example samplesheet and JSON schema for a metadata value
• 3. Parameters
• 4. Output
  • 4.1. Files
  • 4.2. IRIDA Next JSON
    • 4.2.1. Minimal example
    • 4.2.2. Complete example
    • 4.2.3. Output files
      • 4.2.3.1. global
      • 4.2.3.2. samples
    • 4.2.4. Metadata
    • 4.2.5. Creating the IRIDA Next JSON integration file
• 5. Modules
  • 5.1. nf-core modules
  • 5.2. Local modules
    • 5.2.1. Module software requirements
    • 5.2.2 Configuring Module Software with Private or Alternate Container Registries
      • 5.2.2.1 Example: Overriding Container Registries with the container Directive
      • 5.2.2.2 Managing Optional Container Registry Overrides
• 6. Resource requirements
  • 6.1. Process resource label
    • 6.1.1. Accepted resource labels
  • 6.2. Tuning resource limits with parameters
• 7. Testing
  • 7.1. Nextflow test profile
  • 7.2. Unit/integration tests
• 8. Executing pipelines
  • 8.1. Standalone execution
  • 8.2. Execution via GA4GE WES
  • 8.3. Running pipelines in a local instance of IRIDA Next
• 9. Resources
  • 9.1. Pipeline development tutorial
  • 9.2. List of pipelines
  • 9.3. Other resources
• 10. Future pipeline integration ideas
• 11. Security practices
  • 11.1. Automated repository scanning (recommendations)
• 12. Publishing guidelines
• 13. Legal

1. Starting development

IRIDA Next pipelines attempt to follow as much as possible the nf-core guidelines for development pipelines. As a preliminatry first step prior to developing pipelines in IRIDA Next, please read through the following documentation:

• Nf-core guidelines
• Nf-core adding a new pipeline

While we do follow nf-core as closely as possible, there are a number of differences or additions we make for integrating pipelines into IRIDA Next. Hence, some of the nf-core requirements may not apply if you do not intend to contribute the pipeline to nf-core.

1.1. Developing a pipeline to contribute to nf-core

If you intend to develop a pipeline to contribute to nf-core, please follow all the nf-core requirements in addition to the requirements listed in this guide.

1.2. Developing a pipeline which won't be contributed to nf-core

If, instead, the pipeline won't be contributed to nf-core, then only the following set of nf-core requirements would need to be followed, along with the IRIDA Next requirements sections below.

1.2.1. Requirements from nf-core

• Nextflow: Workflows must be built using Nextflow.
• Workflow name: Names should be lower case and without punctuation.
• Use the template: All nf-core pipelines must be built using the nf-core template.
• Software license: Pipelines must open source, released with the MIT license.
• Docker support: Software must be bundled using Docker and versioned.
• Continuous integration testing: Pipelines must run CI tests.
• Semantic versioning: Pipelines must use stable release tags.
• Standardised parameters: Strive to have standardised usage.
• Single command: Pipelines should run in a single command.
• Keywords: Excellent documentation and GitHub repository keywords.
• Pass lint tests: The pipeline must not have any failures in the nf-core lint tests.
  • Note: It is possible to ignore some of the default nf-core linting with a lint configuration file.
• Credits and Acknowledgements: Pipelines must properly acknowledge prior work.
• Minimum inputs: Pipelines should be able to run with as little input as possible.

The following are modified nf-core requirements:

• Bundled documentation: Pipeline documentation should be provided within the README.md, docs/usage.md and docs/output.md files. Note that documentation does not need to be hosted on the nf-core website (unless the pipeline is meant to be contributed to nf-core).
• Use nf-core git branches: Use main (instead of master), dev and TEMPLATE.

The following are additional requirements for using nf-core components outside of nf-core.

• Acknowledging nf-core: Please add an acknowledgement statement to the pipeline README.md for nf-core.
• Removing nf-core only branding: Please remove any nf-core only branding from your repository.

1.3. Requirements for integrating with IRIDA Next

Whether the pipeline you develop is intended to be contributed to nf-core or not, the following are additional requirements for integrating a pipeline with IRIDA Next. These may overlap with existing nf-core and Nextflow standards.

• Input data using samplesheet and schema: Please use a CSV samplesheet for referencing input data, where the structure of the CSV file is described in an assets/schema_input.json file. The JSON schema file will be used to render an interface in IRIDA Next for selecting and sending input data to be executed by a pipeline.
• Parameters described in nextflow.config and nextflow_schema.json: Please describe parameters for the pipeline in the nextflow.config and nextflow_schema.json file. The JSON schema file will be used by IRIDA Next to render an interface for selecting parameters to pass to the pipeline.
• Output a JSON file describing data to load into IRIDA Next: Please write an output/iridanext.output.json.gz file which describes the data to be loaded and stored within IRIDA Next (both files and metadata).
• Describe resource requirements for each module: Please describe resource requirements using only the labels described in the above link.

1.4. Other requirements

The following are other requirements for developing pipelines.

• Software security and vulnerabilities: Please maintain good software security practices with pipeline code to be published. In particular, never store plain text passwords, API keys, ssh keys, or otherwise confidential information in software/pipelines which are distributed publicly. You may consider referring to the GitHub Code Security documentation for securing your repository.

2. Input

Input for pipelines follows the nfcore parameters specification. In particular, input data will be passed via a CSV file, where each row represents data that can be processed independently through the pipeline (commonly a sample). The input CSV file is passed using the --input parameter to a pipeline (e.g., --input samples.csv).

Within Nextflow, this --input samples.csv file is converted into a channel where each element within the channel corresponds to a single row of the input CSV file (e.g., each element in the channel contains the sample identifier and the input fastq files). This conversion from the input CSV file to a channel of tuples is performed using the nf-validation fromSamplesheet function.

Note: Commonly this --input CSV file is referred to as a samplesheet since it's often used to store data associated with a single sample, but it could be used to store other types of data as well.

2.1. Default samplesheet

The default samplesheet looks like:

sample	fastq_1	fastq_2
SampleA	file_1.fq.gz	file_2.fq.gz

2.2. Types of input data

There are three types of columns within a samplesheet used for input: the sample identifier, files, and metadata.

2.2.1. Sample identifier

The sample identifier is used to match up data from the --input samplesheet.csv file to the individual samples this data is associated with. IRIDA Next expects the sample identifier to be stored within the sample column in the samplesheet. This identifier will be used to associate output files and metadata from pipelines with the respective sample (see the 4. Output section for details).

2.2.2. Input files

An input file consists of some data stored within a file external to the CSV file passed to --input. Commonly this would be sequence reads (and would be indicated with the fastq_1 and fastq_2 columns in the CSV file above). However, this could be any data intended to be delivered to pipeline steps via a channel (e.g., bam files of aligned reads, fasta files of assembled genomes, csv files of allelic profiles).

2.2.2.1. Example sample input files

sample	fastq_1 (an input file)	fastq_2 (another input file)	assembly (another input file)
SampleA	file_1.fq.gz	file_2.fq.gz
SampleB			assembly.fa.gz

2.2.3. Metadata input types

A metadata input type is something that can be represented within a cell in a CSV file without reference to some external file (e.g., a Number or a String). In integration with IRIDA Next, these will be derived from contextual metadata of a sample (e.g., Organism).

2.2.3.1. Example metadata input types

sample	organism (metadata input)	fastq_1	fastq_2
SampleA	Salmonella enterica	file_1.fq.gz	file_2.fq.gz

2.3. Defining the samplesheet structure for a pipeline

The structure of an input samplesheet for a pipeline (i.e., column names and data types) is defined within a JSON Schema file found within the assets/schema_input.json file of a pipeline. This JSON Schema file is used by the nf-validation plugin in Nextflow. This plugin defines the keywords used to add new columns to the input samplesheet, documented in the nf-validation samplesheet schema specification. This JSON schema file also accepts the definition of rules for validating the data passed within the samplesheet (e.g., if a column requires only numeric data, this can be defined within the JSON schema).

By default, the samplesheet schema assumes you have three input columns: sample, fastq_1, and fastq_2. Single-end data is handled by leaving fastq_2 blank.

For custom types of data for a pipeline, please modify this JSON schema by modifying the assets/schema_input.json file. Also, make sure your pipeline is configured to use nf-validation fromSamplesheet to create a channel of input data for your pipeline from an input CSV file that follows this JSON schema.

2.3.1. Defining sample identifiers in IRIDA Next

Sample identifiers from IRIDA Next are placed into an input samplesheet.csv for a Nextflow pipeline in the sample column. That is, if the expected samplesheet structure is as below:

samplesheet.csv

sample
SampleA
SampleB

Then the following should be defined in the assets/schema_input.json:

assets/schema_input.json

{
  // ...
  "items": {
    "type": "object",
    "properties": {
      "sample": {
        "type": "string",
        "pattern": "^\\S+$",
        "meta": ["id"]
      }
    }
  }
}

Here, "meta": ["id"] is used internally by Nextflow and the nf-validation plugin to map the identifier in the sample column in the samplesheet.csv to a meta.id value, which is expected to be passed alongside any input data in a channel for Nextflow/nf-core processes. See the nf-validation Sample sheet keys documentation for more details.

2.3.2. Defining selection of input files in IRIDA Next

In order to configure a pipeline to select input files from IRIDA Next, please make sure the samplesheet assets/schema_input.json for the column contains the following keywords:

• "format": "file-type": This is used by IRIDA Next to indicate that a particular column/field in the samplesheet should be a file associated with a sample. This is also by the nf-validation plugin in Nextflow to validate the respective column as containing a file path and check for existence of this file. See the nf-validation schema specification for more details.
• "pattern": ".*.f(ast)?q(.gz)?$": This is used by IRIDA Next to constrain what sorts of files attached to a sample can be used as input to this column (e.g., reads in fastq format, assemblies in fasta format, etc). The value here is a regular expression for a file path. On the Nextflow nf-validation side, this is used to also validate the input paths in the samplesheet are as expected (i.e., they match the defined regular expression).
• fastq_1 and fastq_2 for paired/single-end input fastqs: The samplesheet column names fastq_1 and fastq_2 should be used to indicate input paired/single-end fastq files. Using these column names will mean IRIDA Next will make sure the pairs of reads are properly paired (i.e., if there's multiple read sets attached to a sample it will guarantee that only pairs of reads associated with each other will be allowed as input to the pipeline).
• Other input files (other than paired/single-end fastqs) can use any other column name.

2.3.2.1. Example samplesheet and JSON schema for an assembly

As an example, if a pipeline is intended to take as input an assembled genome (in fasta format) for each sample, the following would be how the samplesheet could be defined:

samplesheet.csv

sample	assembly
SAMPLE1	s1.fasta.gz
SAMPLE2	s2.fasta.gz

The following would be an example structure found within the assets/schema_input.json file:

assets/schema_input.json

{
  // ...
  "items": {
    "type": "object",
    "properties": {
      "sample": {
        "type": "string",
        "pattern": "^\\S+$",
        "meta": ["id"]
      },
      "assembly": {
        "type": "string",
        "format": "file-path",
        "pattern": "^\\S+\\.f(ast)?a(\\.gz)?$"
      }
    }
  }
}

2.3.3. Defining selection of input metadata in IRIDA Next

In order to configure a pipeline to select input metadata from IRIDA Next, please make sure the samplesheet assets/schema_input.json for the column contains does not contain "format": "file-path". That is:

• Does not contain "format": "file-path": This is the key part of defining selection of metadata for IRIDA Next (the assumption being anything that is not a file is metadata).
• Does not use column names sample, fastq_1, or fastq_2: These are reserved column names in a samplesheet for IRIDA Next.
• The column name in samplesheet should match metadata field: IRIDA Next will attempt to automatically select metadata for a pipeline based on the column name in the input samplesheet for a pipeline. That is, if the samplesheet names a column organism, then IRIDA Next will attempt to auto-populate the organism metadata field from the sample data when running the pipline. Users have the option in IRIDA Next to override the selected metadata key when running a pipeline (e.g., if the metadata field in IRIDA Next to use is named species instead).

2.3.3.1. Example samplesheet and JSON schema for a metadata value

As an example, if a pipeline is intended to take as input a metadata value, the following would be how the samplesheet could be defined:

samplesheet.csv

sample	organism
SAMPLE1	Salmonella
SAMPLE2	Salmonella

The following would be an example structure found within the assets/schema_input.json file:

assets/schema_input.json

{
  // ...
  "items": {
    "type": "object",
    "properties": {
      "sample": {
        "type": "string",
        "pattern": "^\\S+$",
        "meta": ["id"]
      },
      "organism": {
        "type": "string"
      }
    }
  }
}

IRIDA Next will attempt to autopopulate the organism column in the samplesheet from the oranism metadata field for samples, but provides a mechanism to override the selected metadata field for samples.

3. Parameters

Parameters within a pipeline are defined in the nextflow.config file (under the params scope), and then the nextflow_schema.json file is updated to contain the parameters by running nf-core schema build. This is described in the nf-core contributing to pipelines documentation.

As an example, the following is a subset of parameters from the iridanext-example-nf pipeline.

params {
    input                      = null
    genome                     = 'hg38'
    igenomes_base              = 's3://ngi-igenomes/igenomes'
    igenomes_ignore            = false
}

The full set of parameters can be found in https://github.com/phac-nml/iridanext-example-nf/blob/main/nextflow.config.

4. Output

4.1. Files

Output files will be placed into a directory specified by the --outdir parameter. For examples given below we assume --outdir output was used.

4.2. IRIDA Next JSON

An output/iridanext.output.json.gz file will be written, which specifies the files and metadata to store into IRIDA Next.

4.2.1. Minimal example

A minimal example of the iridanext.output.json.gz is:

output/iridanext.output.json.gz:

{
  "files": {
    "global": [],
    "samples": {}
  },
  "metadata": {
    "samples": {}
  }
}

This describes the minimal keys and structure for the JSON file.

4.2.2. Complete example

A complete example of this file with more data filled in is as below.

output/iridanext.output.json.gz:

{
  "files": {
    "global": [{ "path": "summary.txt" }],
    "samples": {
      "SampleA": [
        { "path": "assembly/assembly.fa.gz" },
        { "path": "assembly/annotation.gbk.gz" }
      ]
    }
  },
  "metadata": {
    "samples": {
      "SampleA": {
        "organism": "Salmonella enterica",

        "amr.1.gene": "x",
        "amr.1.start": 1234,
        "amr.1.end": 5678,

        "amr.2.gene": "y",
        "amr.2.start": 1,
        "amr.2.end": 2
      }
    }
  }
}

4.2.3. Output files

The files section lists the files to store within IRIDA Next (files not listed here will not be stored). This consists of two different sections:

4.2.3.1. global

The global section defines a list of files associated with the pipeline globally. Each file entry is a JSON object with a single key path defining the path to the file (relative to the output directory). More keys may be added in the future.

"global": [
    {"path": "summary.txt"}
]

4.2.3.2. samples

The samples section defines a list of files to be associated with each sample individual (e.g., genome assemblies). Each sample identifier should correspond to the same identifer specified in the --input samplesheet. Each file entry is a JSON object with a single key path defining the path to the file (relative to the output directory).

"samples": {
    "SampleA": [
        {"path": "assembly/assembly.fa.gz"}
    ]
}

4.2.4. Metadata

The metadata section contains structured data to load within the IRIDA Next database. Currently there is only one type of metadata, samples.

The samples metadata object consists of keys corresponding to sample identifiers used in the original input samplesheet file. The values consist of metadata values to store in IRIDA Next.

As of right now, only simple key-value pairs are supported as metadata. Support for more complicated JSON structures may be added in the future.

"metadata": {
    "samples": {
        "SampleA": {
            "organism": "Salmonella enterica",
            "amr.1.gene": "x",
            "amr.1.start": 1234,
            "amr.1.end": 5678,
        }
    }
}

4.2.5. Creating the IRIDA Next JSON integration file

While the iridanext.output.json.gz integration file can be generated by custom code in a pipeline, in most use cases the expectation is to make use of a Nextflow plugin for this purpose: nf-iridanext.

The nf-iridanext Nextflow plugin can be loaded by a pipeline by adding it to the plugins section within the nextflow.config.

plugins {
    id '[email protected]'
}

This adds support for a custom iridanext scoped block as part of the Nextflow configuration that controls the generation of an IRIDA Next JSON integration file.

The iridanext configuration values can be added anywhere within the Nextflow configuration files. One way to organize the configuration is to create a separate conf/iridanext.config file in the pipeline with the IRIDA Next configuration, and include this file in the nextflow.config.

nextflow.config

includeConfig 'conf/iridanext.config'

conf/iridanext.config

iridanext {
    enabled = true
    output {
        path = "${params.outdir}/iridanext.output.json.gz"
    }
}

When setup this way, the pipeline will generate an iridanext.output.json.gz file within the passed --outdir parameter with the following minimal contents:

{
  "files": {
    "global": [],
    "samples": {}
  },
  "metadata": {
    "samples": {}
  }
}

Please see the nf-iridanext documentation for more details on integrating output files and metadata produced from the pipeline into this IRIDA Next integration file.

5. Modules

Modules in Nextflow are scripts that can contain functions, processes, or workflows.

5.1. nf-core modules

Nf-core contains a number of modules for processes to execute different bioinformatics tools (nf-core modules). Where possible, please use nf-core modules within your pipeline. These can be installed via running the following command from the nf-core tools.

nf-core modules install [NAME]

5.2. Local modules

If it is not possible to use existing nf-core modules, you can create your own modules local to your pipeline in the modules/local/ directory. Please see the nf-core contributing modules documentation for expectations on how modules should behave. Local modules won't need to follow all of these guidelines (such as uploading to the list of nf-core approved modules), but the behaviour of inputs, parameters, and outputs of a process should be followed as closely as possible.

5.2.1. Module software requirements

Each module should define its software requirements as a singularity/docker container. This requires using the container keyword in the process. For example:

container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
    'https://depot.galaxyproject.org/singularity/fastqc:0.11.9--0' :
    'biocontainers/fastqc:0.11.9--0' }"

Since IRIDA Next Nextflow pipelines aim to follow nf-core standards, the default container registry should be set to the publicly available quay.io registry in the nextflow.config file using the .registry parameters. For example:

docker.registry =  'quay.io'

For more information, see the Nextflow containers documentation and the nf-core modules software requirements guide.

5.2.2 Configuring Module Software with Private or Alternate Container Registries

To configure a private or alternate container registry, the standard practice is to set the .registry option to point to the desired registry in the nextflow.config file. For example:

docker.registry = 'container-registry.com'

This setting automatically prefixes each container name with the specified registry URL. In the above example, biocontainers/fastqc:0.11.9--0 is transformed into container-registry.com/biocontainers/fastqc:0.11.9--0, ensuring that the container is downloaded from container-registry.com.

In IRIDA Next Nextflow pipelines, we have chosen to adhere to nf-core standards by designating quay.io as the default container registry. This setup ensures that containers, such as biocontainers/fastqc:0.11.9--0, are automatically referenced as quay.io/biocontainers/fastqc:0.11.9--0. While this configuration simplifies container management, it can be restrictive when the pipeline requires containers from multiple registries, such as both quay.io and docker.io, since only one default registry can be configured at a time. Introducing flexibility in sourcing containers from different registries becomes valuable in situations where specific containers are only available in certain registries, or where there are preferences or restrictions on registry use.

To address this, we use process.ext.override_configured_container_registry to manage the selection of container registries dynamically by adding the following line to the nextflow.config file:

nextflow.config

// Override the default Docker registry when required
process.ext.override_configured_container_registry = true

This custom extension allows us to define configurations beyond the standard Nextflow syntax, tailored to the needs of specific processes in the workflow. When a process runs, the value of process.ext.override_configured_container_registry is accessed within the container directive as task.ext.override_configured_container_registry. The task context refers to the specific instance of a process execution, encompassing all relevant attributes and extensions for that instance. This setup ensures that any custom settings, such as overriding the default container registries, are correctly applied during the execution of the pipeline.

5.2.2.1 Example: Overriding Container Registries with the container Directive

To specify an alternative registry for a process, include a conditional check with task.ext.override_configured_container_registry in the nested ternary operator of the container directive. For example:

container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
    'https://depot.galaxyproject.org/singularity/fastqc:0.11.9--0' :
    task.ext.override_configured_container_registry != false ?
    'docker.io/biocontainers/fastqc:v0.11.9_cv8' :
    'biocontainers/fastqc:0.11.9--0' }"

In this example, setting the task.ext.override_configured_container_registry in the container directive ensures the default Docker registry (quay.io) is overridden, allowing the process to use the DockerHub container registry (docker.io) instead.

By using != false in the condition, we ensure that the process will follow the alternative registry path whenever task.ext.override_configured_container_registry is set to any value other than false (including true or null). This approach provides flexibility by allowing different registry configurations to be applied without needing explicit overrides for every scenario.

5.2.2.2 Managing Optional Container Registry Overrides

To provide greater flexibility in managing the process.ext.override_configured_container_registry setting, users can create an optional configuration file separate from the nextflow.config to define whether an alternative container registry can be used in the pipeline. This file can then be included into the Nextflow pipeline by using the -c option on the commandline.

For example, you can create a configuration file named azure_configuration.config to set both the default container registry (in this case the docker.registry) and the process.ext.override_configured_container_registry setting, which controls whether an alternative container registry can be specified in the Nextflow pipeline.

azure_configuration.config

// Azure-specific private registry to pull containers from
docker.registry = "private-registry.com"

// Azure-specific configuration for overriding the default container registry
process.ext.override_configured_container_registry = false

Include this configuration file in your pipeline with the following command:

nextflow run ... -c azure_configuration.config

This method allows for dynamic adjustment of container settings, enabling flexible selection of container registries tailored to various environments or specific requirements, such as those used in Azure.

6. Resource requirements

To define computational resource requirements for each process, we will follow the nf-core resource standards as much as possible, where resources are adjusted by a label in each process of a pipeline. The pipeline developer will be responsible for setting an appropriate label in each process of the NextFlow pipeline to appropriatly define required resources. In addition, the developer is responsible for tuning any resources defined in the config/base.config file, as described in the nf-core tuning workflow resources documentation.

6.1. Process resource label

The pipeline developer will add a label to each process of a NextFlow pipeline to adjust resources. For example:

process name {
    label 'process_single'

    // ...
}

6.1.1. Accepted resource labels

The following labels will be accepted:

• proccess_single
• process_low
• process_medium
• process_high
• process_very_high: This label is an addition over those provided by nf-core to be used for situations where a process needs a lot of resources beyond process_high.

6.2. Tuning resource limits with parameters

Nf-core provides the capability to adjust the maximum resources given to processes in a pipeline using parameters: --max_cpus, --max_memory, and --max_time. Pipeline developers will be responsible for making sure thse parameters are available in the nextflow.config file. See the nf-core max resources documentation for more details.

7. Testing

Nextflow pipelines should include test suites, linting, and execution of these tests automatically on merging of new code (i.e., continuous integration). These files should be included automatically if the pipeline is generated using the nf-core create command. A description of these different tests are included in the nf-core contributing documentation.

7.1. Nextflow test profile

In particular, a test profile should be setup in Nextflow so that the pipeline can be executed with nextflow [NAME] -profile test,docker.

Parameters for the test profile should be specified in the conf/test.config file. This can reference data stored elsewhere (e.g., on GitHub). For example, see the test.config for the IRIDA Next sample pipeline https://github.com/phac-nml/iridanext-example-nf/blob/main/conf/test.config.

7.2. Unit/integration tests

It is also expected to include with each pipeline a set of unit/integration tests. These can be implemented by using nf-test.

Documentation is provided on the nf-test website, but the quick way to get started is to first install nf-test (e.g., via conda):

conda create --name nextflow nextflow nf-test

Then, run the following to create the necessary files for nf-test tests:

nf-test init

The nf-core templates for pipelines should include a GitHub actions task for running nf-test on pull-requests. Please refer to the nf-test documentation for more details.

8. Executing pipelines

8.1. Standalone execution

Pipelines should be configured so that they can be executed by:

nextflow run [NAME] --input inputsheet.csv --outdir output [OTHER PARAMETERS]

Where inputsheet.csv is the CSV file containing samples (or other input data) and output is the directory containing output data. Other parameters can be included, though reasonable defaults should be set so the number of required parameters is as minimal as possible.

8.2. Execution via GA4GE WES

IRIDA Next will make use of the GA4GH Workflow Execution Service API for executing pipelines. This will require making a POST request to RunWorkflow with JSON that looks similar to the following:

{
  "workflow_params": {
    "--input": "https://url-to-input.csv",
    "-r": "REVISION",
    "[PARAMETERS]": "[PARAMETER VALUES]"
  },
  "workflow_type": "DSL2",
  "workflow_type_version": "22.10.7",
  "tags": {
    "createdBy": "Test",
    "group": "Test"
  },
  "workflow_engine_parameters": {
    "engine": "nextflow",
    "execute_loc": "azure"
  },
  "workflow_url": "https://github.com/phac-nml/iridanext-example-nf",
  "workflow_attachment": ""
}

Here, parameters are specified by key/value pairs under workflow_params and the location of the workflow code is given in workflow_url. See WES RunWorkflow for more details.

8.3. Running pipelines in a local instance of IRIDA Next

For detailed instructions and configuration options, please refer to the the guide for setting up a local instance of IRIDA Next:

IRIDANext_LocalGuide.md

The guide includes step-by-step instructions for installing required dependencies, setting up the development environment, and configuring the necessary services.

• Summary of Installation Steps
  • Install asdf: A tool version manager for managing dependencies.
  • Install Required Plugins: Node.js, pnpm, Ruby, and Postgres using asdf.
  • Setup and Start Postgres Server: Create a new role and configure the development database.
  • Configure Environment: Set up credentials and active storage for IRIDA Next.
  • Setup Sapporo: Install and configure Sapporo for managing GA4GH WES instances.
  • Register Pipelines: Add and register Nextflow pipelines in IRIDA Next.
  • Run Local Instance: Start the web application and access it via a web browser.

9. Resources

9.1. Pipeline development tutorial

A tutorial on developing a pipeline is available at IRIDA Next nf-core pipeline tutorial.

9.2. List of pipelines

An example pipeline that conforms to these standards is available at https://github.com/phac-nml/iridanext-example-nf.

Other pipelines are listed at https://github.com/phac-nml/nf-pipelines.

9.3. Other resources

• PHA4GE proposed pipeline standards: Proposed pipeline standards from the PHA4GE Bioinformatics Pipelines & Visualization Working Group
• Government of Canada: Guide for Publishing Open Source Code
• nfcore pipeline schema
• nfcore parameters
• GA4GH Workflow Execution Service

10. Future pipeline integration ideas

Additional ideas for integrating pipelines in the future can be found in the IRIDA Next pipeline integration ideas document.

11. Security practices

Please maintain good software security practices with pipeline code to be published. In particular, never store plain text passwords, API keys, ssh keys, or otherwise confidential information in software/pipelines which are distributed publicly.

11.1. Automated repository scanning (recommendations)

There exist automated tools to aid in maintining secure code and identifying potential issues. The GitHub Securing your repository documentation provides detailed information on how to secure your GitHub repository, which includes automated scanning. In particular, some recommendations include:

• Enable secret scanning: This will scan code and issues for patterns potentially matching secrets (e.g., API keys) that should not be made public and provide alerts for anything that was found.
• Enable code scanning: GitHub provides CodeQL for scanning code for potential vulnerabilities or errors in your code and providing recommended fixes. While this does not support the Nextflow language, other programming languages (e.g., Python) are supported.
• Enable dependabot: Dependabot scans for and provides alerts on insecure dependencies of your software. While Nextflow is not directly supported by github, this might be a service of use for other repositories or other code and dependencies managed within a pipeline.

Enabling these services can be done directly on a GitHub repository in the Settings > Code security and analysis section.

Once enabled, security alerts appear in the Security tab on GitHub.

For more security recommendations, please see the PHA4GE pipeline guidelines.

12. Publishing guidelines

Our intention is to follow, as much as possible, the standards and practices set out by nf-core. However, we leave it as optional to actually publish pipelines/modules/subworkflows with the official nf-core repositories. We would encourage this where it makes sense in order to support re-use and giving back to the community (please refer to the nf-core publishing requirements for the guidelines in this case). However, it is perfectly acceptible to publish pipelines/modules/subworkflows in separate Git repositories outside of nf-core. Please see the if the guidelines don't fit and using nf-core components outside of nf-core sections of the nf-core documentation for more information on this scenario and other locations to list your pipeline.

13. Legal

Copyright 2023 Government of Canada

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this work except in compliance with the License. You may obtain a copy of the License at:

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.