I recently prepared a small Docker training for our research group in order to highlight our new CobraPy Docker container.
I think that Docker is a pretty great technology for Computational and Systems Biology. Most researchers in those areas do a lot of programming, but are actually neither trained nor paid for the software development. Due to this, scientists normally do not suffer from the NIH syndrome (not related to the funding agency, see here), but rather combine a lot of different programs and programming languages into a single pipeline to achieve their goals. In general, I am not a huge fan of that, because this work flow is inherently difficult to maintain (should I write a post about that?). However, reality is that scientific software is a lot of pipelines and Docker is a nice tool to ensure reproducibility. Basically, it makes it easy to deploy an isolated environment containing your entire pipeline with the required software in the required versions.
In my opinion, all projects that are not a single language package (not Python/R packages for instance) should be bundled into a Docker container upon publication to give the community a way to get your pipeline with a single command and reproduce your work. Additionally, Dockers automated builds can help you greatly to maintain your pipeline and easily test new software versions.
So if that spiked some interest you can use the Github link on the bottom of the page to get to my Docker Mini Training which will show you how to wreck a Debian system and how to deploy your very own Darth Vader Cow app.