Continuous Integration with Travis

This document summarizes the steps involved in setting up a continuous integration suite with Travis for our pipelines.

Getting Started

Login with your gitHub account details on Add the “Multiscale Genomics” to your organization to gain access to its repositories. (you would have to send a request to be added to this organization).

Follow the onscreen instructions on

Flick the repository switch to "on".
Add .travis.yml to your repository.
Sync your travis account to your GitHub account.
Pushing to Git will trigger the first Travis build after adding the above file.

Making .travis.yml File

To your added .travis.yml file in your GitHub repository include:


libtbb did not seem to be installing correctly when put in “apt: packages: “. It has therefore been done with sudo in “install:”

  • Setup all symlinks in “before_script: “
  • Change execution permissions on shims folder and
  • Add the shims folder path to your $PATH
  • Include to your “script” to be executed

Making File

Include all test scripts to be tested with pytest to your file. As iNPS does not work with Python < 3, a conditional check is present to ensure that it does not run unless Travis is running Python 3

Running Docker container

To run docker within Travis, it has been included within the “services” in the .yml file. Every time travis runs docker it will pull the latest publicly available image from DockerHub and run it. This gets done from within the shims files. The pre-built container can also be pulled from :

using command :

docker pull multiscalegenomics/mgprocessfastq:testdocker

For more details on the docker container, please refer to :

Setting up Shims

Libmaus2 and Biobambam2 have had to be installed within a docker container, as they were causing Travis to time out. As the container is non-interactable while on Travis and is not hosting live server. An intermediate layer to access the contents within docker from travis has been introduced in the form of shims. To make these files, take the list of all biobambam2 modules and construct bash script files with the following command for each of the modules :

exec docker run -it  multiscalegenomics/mgprocessfastq:biobambamimage ~/lib/biobambam2/bin/biobambam_module_name $@

The .travis.yml file used for testing the mg-process-fastq pipelines can be found at :