Introduction to Reproducibility in Cancer Informatics
This course was created from this GitHub template.
You can see the rendered course material here: https://jhudatascience.org/Reproducibility_in_Cancer_Informatics
If you would like to contribute to this course material, take a look at the getting started GitHub wiki pages.
About this course
This course introduces the concepts of reproducibility and replicability in the context of cancer informatics. It is the first course in a two part course on reproducibility (See the advanced course here). It uses hands-on exercises to demonstrate in practical terms how to increase the reproducibility of data analyses. The course also introduces tools relevant to reproducibility including analysis notebooks, package management, git and GitHub.
Learning Objectives
This course will teach learners to:
- Understand the fundamental concepts of reproducibility vs replicability
- Write a basic analysis that could be passed to another person/computer and be re-ran to obtain the same output
- Write analysis notebook that adheres to best practices for reproducibility and readability
- Create a reproducible project on GitHub that could be re-run by someone else
Encountering problems?
If you are encountering any problems with this course, please file a GitHub issue or contact us at {Some email or web address with a contact form}.
All materials in this course are licensed CC-BY and can be repurposed freely with attribution.
About the chapter example files
Each chapter has its own zip file to be downloaded as a learner is going through a particular chapter.
There are the Python and R versions. For example, for chapter 4 there are:
python-examples/python-heatmap-chapt-4
r-examples/r-heatmap-chapt-4
These files are zipped up by a GitHub action so they are ready for easy downloading by the learner. The user will download these according to the chapter and follow along with the chapter to make them more reproducible and eventually hopefully have something that looks like the "final" reproducible example versions.
This is the URL pattern they can find the chapter files at:
For Python:
https://github.com/jhudsl/Reproducibility_in_Cancer_Informatics/raw/main/chapter-zips/python-heatmap-chapt-4.zip
For R:
https://github.com/jhudsl/Reproducibility_in_Cancer_Informatics/raw/main/chapter-zips/r-heatmap-chapt-4.zip
Obtaining the "final" versions of the example reproducible analyses
Both the "final" versions of the example analyses have their own repositories that are submodules of this one (located in their respective directories with the less reproducible versions of them in the r-examples and python-examples directories).
Final here is in quotes because we may continue to make improvements to this notebook too -- this course tries to emphasize that work on data analyses should be iterative and we never have to say we're done with an analysis if we find other ways it can be improved!
- https://github.com/jhudsl/reproducible-python-example
- https://github.com/jhudsl/reproducible-r-example
Running the R docker image:
With your current directory being the top of this repository, you can do:
cd r-examples/reproducible-r-example
docker build -f docker/Dockerfile . -t jhudsl/reproducible-r
docker run -it -v $PWD:/home/rstudio -e PASSWORD=password -p 8787:8787 jhudsl/reproducible-r
Then, in the browser of your choice, navigate to localhost:8787 ; using rstudio as your username and password as your password (or whatever you choose for your password in the command above). This docker image has the renv included in it.
Running the Python docker image:
With your current directory being the top of this repository, you can do:
cd python-examples/reproducible-python-example
docker build -f docker/Dockerfile . -t jhudsl/reproducible-python
docker run --rm -v $(pwd):/home/jovyan/work -e JUPYTER_ENABLE_LAB=yes -p 8888:8888 jhudsl/reproducible-python
Then, in the browser of your choice, navigate to the port that the output tells you. This docker image will automatically have your conda environment set up and working.