Data Management

Session Materials

Class examples and Reproducibility resources

Contact information

Presenter: April Clyburne-Sherin
Email: april@codeocean.com \ Slidedeck: MSK Reproducible Data Management Slidedeck

Links to the recommended resources

Good enough practices in scientific computing. Wilson G, Bryan J, Cranston K, Kitzes J, Nederbragt L, et al. (2017) Good enough practices in scientific computing. PLOS Computational Biology 13(6): e1005510. https://doi.org/10.1371/journal.pcbi.1005510

Ten Simple Rules for Reproducible Computational Research. Sandve GK, Nekrutenko A, Taylor J, Hovig E (2013) Ten Simple Rules for Reproducible Computational Research. PLOS Computational Biology 9(10): e1003285. https://doi.org/10.1371/journal.pcbi.1003285

The Practice of Reproducible Research: Case Studies and Lessons from the Data-Intensive Sciences. Kitzes, J., Turek, D., & Deniz, F. (Eds.). (2018). Oakland, CA: University of California Press. https://www.gitbook.com/book/bids/the-practice-of-reproducible-research/details

Jupyter Notebooks and reproducible data science. Woodbridge M, Sanz D, Mietchen D, & Mounce R (2017).

---

Installations & Prerequisites

• GitHub account is recommended.
• No installations are necessary.

Session description

Organization

Data preparation & collection:

• Create a data management plan.
  • Resource: DMPTool: https://dmptool.org/

Centralize your project in a repository:

• Create one repository that holds all related research files: data, code, notebooks, documentation, etc.

Create directory structure:

• Organize your research to separate code + notebooks from data.
• Save results explicitly.
• Resource on reproducible organization:
  • Karl Broman: http://kbroman.org/steps2rr/pages/organize.html

Documentation

Specify run environment:

• Configure the run environment for your code.
  • Example: Base Environment: Python 3 with Anaconda

Specify dependencies:

• Specify your packages and dependencies with versions.
  • Example: conda installer: jupyter, numpy, pandas
  • pip freeze > ../code/requirements.txt
  • conda list –export > ../code/requirements.txt
• Configure using container technology to capture your dependencies.
• Resource on documenting dependencies:
  • Binder: http://mybinder.readthedocs.io/en/latest/using.html#preparing-a-repository-for-binder

Create a project level README:

• Create a README.txt or README.md.
• Resource on making a great README file:
  • Cornell (includes a template): https://data.research.cornell.edu/content/readme
• Documenting packages in your README:
  • AJPS Replication Package: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/EZSJ1S
• Resource on using markdown:
  • GitHub: https://github.com/adam-p/markdown-here/wiki/Markdown-Cheatsheet

Create metadata:

• Create a data dictionary.
• Resources on making a great data dictionary:
  • DataONE: https://www.dataone.org/best-practices/create-data-dictionary
  • McGill Codebook Cookbook: http://www.medicine.mcgill.ca/epidemiology/joseph/pbelisle/CodebookCookbook.html
  • UPenn: https://guides.library.upenn.edu/datamgmt/documentation
  • Karl Broman: http://kbroman.org/dataorg/pages/dictionary.html
• Example codebook:
  • AJPS Replication Package: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/EZSJ1S

Automation

Create a master script:

• Create a master script that executes your notebooks in order.
  • Create a file in the code directory.
  • Name the file “run.sh”.
• Use nbconvert to render your notebook.
  • In your run.sh script, use nbconvert to execute your notebook into the results directory.
• Resource on automation using a driver script: https://www.practicereproducibleresearch.org/core-chapters/3-basic.html

Create relative paths:

• Change absolute paths to relative paths.
• Resource explaining paths:
  • Karl Broman: http://kbroman.org/steps2rr/pages/organize.html

Dissemination

Specify a licence:

• Specify a license for your data and your code + notebooks.
• Resource on choosing a data licence:
  • Digital Curation Center: Digital Curation Centre
• Resource on choosing a code licence:
  • Karl Broman: http://kbroman.org/steps2rr/pages/licenses.html
  • Open Source Initiative: https://opensource.org/licenses

Publish your code and data:

• Share your repository.
  • Obtain a DOI for your repository and use this link throughout your article.
  • Example: Github -> Binder -> Zenodo -> DOI linked in article

Recording

The recording of this session is here