At the Data Modality and Methodology Task Force, we have been developing a lightweight, structured template for starting a one-person analytical project.
Key Features:
Reproducibility – Clear folder structure and easy setup with a virtual environment
Security – Prevents sensitive data and keys from being pushed to GitHub with .gitignore
Simplicity – No complex dependencies, just a clean starting point
How to Use:
Clone the repository and rename it
Set up the virtual environment and install basic packages
Move your data to the data/ folder, install more packages for your need
Start coding!
By sharing your repository, others should be able to reproduce your results as long as they have access to the data.
This is very cool, thanks for sharing @hirotaka ! Do you have any examples of projects that may have (or plan to) use this? Are there any limitations to the types of data modalities this can work with or analytical approaches that this can make use of?
Thank you for your comments, @jgottesman! I’ve been using this template for my own projects for over a year. While all of these projects are currently private, I’ll consider creating a public one based on a test analytical project.
Regarding limitations, this repository includes virtual environment setup for Python but not for R (or other languages). If someone needs full control over R versions and package dependencies, I’m not sure of the best approach. Docker comes to mind, but are there any lightweight alternatives to pyenv or venv for R?
I can confirm the usefulness of this template. This is quite easy to implement and to follow, even for me, someone who always did coding using Cloud environments such as Terra and Google Colab. My project on Useful PPMI Clinical Codes - Code Available - #4 by danieltds is entirely based on this structure, which I beta tested, and it works flawlesly!