We have defined a project lifecycle to make our experiments reproducible.
A project is made of two components:
- the workflow: that is all your code to perform analysis and run experiments.
- the workspace: that is a directory on a cluster or workstation, where all your experiments take place.
We have defined templates, called
cookiecutters, to setup a workflow and a
workspace according to our standards.
Managing your project is a 3 step process:
Develop a workflow
- Create a workflow using
Please, read how to prepare your workflow here.
- Write the code for your workflow.
- Add test data and package required software with Docker.
- Push your workflow to GitHub.
- Create a project workspace using the
Please, read carefully how the workspace is organized here.
- Pull your workflow from GitHub.
- Set your experimental parameters.
- Run your experiments.
- Create a report using the
- Write your report including:
- methods used and/or implemented;
- experimental setup, e.g. parameters, workflow version etc.
- tables and figures