A recipe for extremely reproducible enrichment analysis

By Mark Ziemann and Anusuiya Bora

Computational reproducibility is difficult, especially over longer timeframes.

Using a Docker image allows researchers to keep a stable compute environment with a known version of operating system, R and packages. The downside of Docker is that working with it is typically done with the command line, and requires a relatively high degree of specialised knowledge.

This work is designed to help researchers make their enrichment analyses more reproducible by providing two important things:

• Templates for a functional Docker container with R-based enrichment tools.

• Step-by-step guides to help users customise templates and deploy analyses in their research projects.

Content of this repository

• README.md: this file.

• Dockerfile: this is the build instruction set for the Docker image.

• example.Rmd: this is the Rmarkdown workflow which you should modify to suit your needs.

• manuscript.Rmd: we wrote the whole manuscript in Rmarkdown to prove we could make the whole process reproducible.

• docs: Step-by-step guides to getting your transcriptome analyses done in Docker.

To reproduce this workflow

docker pull mziemann/enrichment_recipe

docker run -it -v ${pwd}:/enrichment_recipe --entrypoint /bin/bash mziemann/enrichment_recipe

Then inside the container, execute the Rmarkdown script.

R -e 'rmarkdown::render("example.Rmd")'

After it completes, exit R and the container. Then copy the report from the container to the working directory.

docker cp $(docker ps -aql):/enrichment_recipe/example.html .

The result you obtain should be identical to this one we generated previously.

Rebuild the docker image

If for some reason you need to rebuild the docker image, use the following:

docker build -t mziemann/enrichment_recipe --no-cache .

Note that the Dockerfile must be in the current directory. The "no cache" option forces docker to repeat the entire build, even if a cached occurrance is saved.

Data and Software Availability

Underlying data

Publicly available data were obtained from Digital Expression Explorer 2 (dee2.io). Expression data from untreated and Aza treated AML3 cells. Accession Number SRP038101. Data are available under the terms of the GNU General Public License v3.0.

Software and Code

• Code repository including template Dockerfile and R Markdown script is available on GitHub (https://github.com/markziemann/enrichment_recipe).

• The example Docker image was deposited to DockerHub (https://hub.docker.com/r/mziemann/enrichment_recipe).

• The code repository and Docker image have been uploaded to Zenodo for long-term archiving (https://zenodo.org/record/8170984).

Availability of other materials

• Protocol: can be found at protocols.io (https://www.protocols.io/view/a-recipe-for-extremely-reproducible-enrichment-ana-j8nlkwpdxl5r/v2).

• Instructional video guides are available on YouTube (https://www.youtube.com/playlist?list=PLAAydBPtqFMXDpLa796q7f7W1HK4t_6Db).