Skip to main content

Contains classes and helpers to build a workflow, and provide options to convert to CWL / WDL

Project description

Janis

GitHub stars Build Status Documentation Status PyPI version codecov Code style: black

Janis is a framework creating specialised, simple workflow definitions that are then transpiled to
Common Workflow Language or Workflow Definition Language.

Documentation is hosted here: https://janis.readthedocs.io/

Introduction

WARNING: this project is work-in-progress and is provided as-is without warranty of any kind. There may be breaking changes committed to this repository without notice.

Janis gives you an API to build computational workflows and will generate a workflow description in CWL and WDL. By using Janis, you get type-safety, portability and reproducibility across all of your execution environments.

Janis requires a Python installation > 3.6, and can be installed through PIP (project page):

# Install janis and the bioinformatics tools
pip3 install janis-pipelines[bioinformatics]  

You can import Janis into your project with:

import janis as j  

Usage

Janis has an API that mirrors the workflow concepts:

  • j.Workflow: A workflow represents the Edges between Input, Step, Output

    • j.Input: An input to a Workflow, has an identifier, a type and a value.
    • j.Step: A step also has an identifier and a Tool (CommandTool or a nested Workflow).
    • j.Output: An output to a workflow has an identifier and is connected to a step.
  • j.CommandTool: A command line style tool that builds it's command through the inputs and arguments.

    • j.ToolInput: An input to a tool, has an identifier, a type and command line options like position, prefix
    • j.ToolArgument: An argument to a tool that cannot be overridden. Has a value and command line options like position and prefix. The value can be a derived type, like an InputSelector or StringFormatter.
    • j.ToolOutput: Output to a tool, has an identifier, a type and a glob.

Example

Further information: Simple Workflow

Below we've constructed a simple example that takes a string input, calls the echo tool and exposes the Echo tool's output as a workflow output.

import janis as j  
from janis.unix.tools.echo import Echo   

w = j.Workflow("workflowId")  

inp = j.Input("inputIdentifier", j.String(), value="my value to print")  
echostep = j.Step("stepIdentifier", Echo())  
outp = j.Output("outputIdentifier")  


w.add_edges([
    (inp, echostep.inp),    # Connect 'inp' to 'echostep'
    (echostep, outp.outp)    # Connect output of 'echostep' to 'out'
])

# Will print the CWL, input file and relevant tools to the console  
w.translate("cwl")  # or "wdl"

We can export a CWL representation to the console using .translate("cwl").

Named inputs and Outputs

Every input and output of a tool is named. In this example, Janis knows that there is only one input and one output of echostep, so can automatically connect these together. You should see a statement in the console that indicates that Janis has automatically made this connection.

[INFO]: The node 'stepIdentifier' was not a fully qualified input of the tool 'Echo', this was automatically corrected (stepIdentifier → stepIdentifier.inp)
[INFO]: The node 'outputIdentifier' under-referenced an output the step 'stepIdentifier' (tool: 'Echo'), this was automatically corrected (stepIdentifier → stepIdentifier.outp)

Included tool definitions and types

Bioinformatics

The Janis framework can be extended to include a suite of Bioinformatics data types and tools. These can be installed with the bioinformatics install extra option.

pip3 install janis-pipelines[bioinformatics]  

Unix

Tool document:

Some basic unix tools have been wrapped and included as part of the base Janis module and are the basis for the examples. You can reference these unix tools through janis.unix.tools.

These can be referenced by janis.bioinformatics or janis_bioinformatics, the latter might be easier due to the way nested python imports work.

More examples

About

Further information: About

This project was produced as part of the Portable Pipelines Project in partnership with:

References:

Through conference or talks, this project has been referenced by the following titles:

  • Walter and Eliza Hall Institute Talk (WEHI) 2019: Portable Pipelines Project: Developing reproducible bioinformatics pipelines with standardised workflow languages
  • Bioinformatics Open Source Conference (BOSC) 2019: Janis: an open source tool to machine generate type-safe CWL and WDL workflows
  • Victorian Cancer Bioinformatics Symposium (VCBS) 2019: Developing portable variant calling pipelines with Janis

Support

Contributions

Further information: Development

This project is work-in-progress and is still in developments. Although we welcome contributions, due to the immature state of this project we recommend raising issues through the Github issues page for Pipeline related issues.

If you find an issue with the tool definitions, please see the relevant issue page:

Information about the project structure and more on contributing can be found within the documentation.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

janis pipelines-0.3.1.tar.gz (58.1 kB view hashes)

Uploaded Source

Built Distribution

janis_pipelines-0.3.1-py3-none-any.whl (88.9 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page