A data science project template

This is a cookiecutter template for data science projects. It sets up stubs for:

• standard folder system with README.md descriptions within each folder
• Luigi build pipeline, build.py, for downloads and batch preprocessing
• Dockerized notebook server for either Jupyter, RStudio, or RStudio with Bioconductor packages pre-installed.
• computational notebook to run your analysis, either notebook.ipynb (if you use Jupyter) or notebook.Rmd (if you prefer RMarkdown).

Docker allows you to completely specify your enviroment so your analysis is fully reproducible. Just make sure to add all dependencies (e.g., apt-get for system packages, conda for Python, CRAN or Bioconductor for R) to the Dockerfile and, if applicable, the install_packages.r script used by the Dockerfile.

You can choose the server platform during initiation of the cookiecutter script, which will then adjust the Dockerfile accordingly.

Get started

From a terminal, change to the directory you want to hold the project folder, and run cookiecutter:

$ cookiecutter https://github.com/austinkeller/datascience-project-template

Fill in the prompts (your name and email, project name and description, and the flavor of notebook platform) and it will be created for you! Be sure to double-check the auto-generated project slug and fix it if necessary–this will be the name of the folder.

Adapted from https://github.com/outlierbio/ob-project-template
Inspired by OutlierBio's guide to organizing data science projects