How I Teach Life Scientists by Using Reproducible and Scalable Learning Environments

Geert van Geest, SIB Swiss Institute of Bioinformatics

Format: Short talk and demo
Take home messages

Docker can be used to develop reproducible learning environments
Containers can easily be hosted on a cloud computing service
The combination of Docker + cloud computing service enables a teacher to create a highly scalable and flexible learning environment

Abstract

While teaching topics in bioinformatics, setting up a computer learning environments can be challenging. Installing and using required software on the computers of the course participants can be tricky because there is usually a wide range of hardware specs, operating systems, versions, permissions, and settings. In addition, some processes that a teacher would like to run during the course might require a specific operating system or computational resources (e.g., high memory or a GPU). Here, I present an approach to host containers on a cloud service that enables the teacher to: (1) develop a reproducible teaching environment with docker, (2) scale calculations to a wide range of computational requirements on a cloud platform. We have applied this approach to several different courses, including the topics ‘single cell transcriptomics’, ‘variant analysis’ and ‘introduction to NGS’. All containers we have used so far are based on either Rstudio server, jupyterLab or VScode images and are therefore easily approachable by the participants through a web browser. The documentation and scripts to apply this method for teaching can be found at:

Resources