de.NBI/CIBI Contributions

de.NBI/CIBI

The 'German Network for Bioinformatics Infrastructure – de.NBI' is a national infrastructure supported by the Federal Ministry of Education and Research providing comprehensive, high-quality bioinformatics services to users in life sciences research and biomedicine.

The 'Center for Integrative Bioinformatics – CIBI' is a de.NBI service center consisting of the three partners KNIME (University of Konstanz), OpenMS (University of Tübingen) and SeqAn (Freie Universität Berlin). CIBI provides KNIME integrations for OpenMS and SeqAn, user and developer trainings and custom KNIME extensions for bioinformatics.

If you use this KNIME extension, please take 5 minutes to give us your feedback using our survey.

Overview

The de.NBI/CIBI KNIME extension contains useful nodes and im- and exporters for working with life science data repositories. This includes nodes for downloading data from repositories such as PubChem, PRIDE and Peptide atlas, as well as menu extensions for importing and exporting workflows from/to myexperiment.org.

Components

The following components are currently part of the extension:

PubChem nodes

Nodes that retrieve data from bio assays and chemical structures of compounds and substances from PubChem via the Power User Gateway.

Other repository downloader nodes

With the PRIDE and Peptide Atlas downloader nodes, it is possible to query the respecitve repository and download the files associated with a study or experiment.

MyExperiment Im- and Exporter

The myexperiment extension enables the user to upload workflows to the myexperiment workflow repository through an easy to use wizard. Additionally it provides the capability to download workflows hosted on myexperiment by specifying the workflow id or URL. 

Docker Exporter

The Docker Exporter allows the creation of Docker images from either single workflows or workflow groups.

To export a workflow or a group, select it in the KNIME Explorer and then click on File -> Export as Docker image in the main menu. A wizard will guide you to the steps of creating the image.

The generated image is based on images that are distributed via Github:

https://github.com/ibisba/knime-base

https://github.com/ibisba/knime-workflow-base

If you choose your own base image, make sure that it is compatible to knime-workflow-base, i.e. has at least the same build arguments.

1. Page: Workflow selection

On the wizard's first page you can select a workflow to export. If a workflow is selected in the KNIME Explorer, this workflow or workflow group should already appear here. Once the correct workflow is selected, proceed to the next page.

2. Page: Export Options

On the second page you have to enter the image tag you want to export the image to, the update sites to use for installing required plugins and the base image. The base image needs to build separately for each version of KNIME. Currently there is no official base image, but you can build one yourself using the attached Dockerfile, where you may have to change the KNIME version it loads (then remove the .txt suffix to use it with docker build).

3. Page: Docker Client Settings

Here you can change the way the client connects to a Docker daemon. Please keep in mind that docker may use a Unix socket, which requires you to run KNIME as administrator.

4. Page: Additional Settings

If you need any additional commands to be run in the workflow's Dockerfile, e.g. for installing additional software, you can enter them here. Prepend each command you want to run in the shell with "RUN". For further information consult the Dockerfile reference.

 

Click "Finish" to start the export process. Once it is finished, you will see a message box showing you the image ID and hints how to run the image.

To run an image that contains a workflow group, use a command like this:

docker run -v /my/local/data/dir:/data mycoolworkflow <relwfpath> workflowvar1=file:/data/iris.csv workflowvar2=file:/data/model.pmml

This commands mounts a local folder in the container under the path /data. The image tag is "mycoolworkflow" and the path to the workflow within the group is <relwfpath>. Here we also pass two workflow variables called workflowvar1 and workflowvar2. To reduce the size of the command, we recommend saving the workflow settings in a file that is mounted into the container and read by the workflow. If you exported a single workflow, simply omit the <relwfpath>.

To list all workflows in an image containing a workflow group, use:

docker run -rm mycoolworkflow --info

or

docker run -rm mycoolworkflow --vars