sjackman
commented
8 years ago - Record URLs from which data was downloaded in a script
- In that same script calculate the SHA256 of the downloaded files right after they're downloaded
- Include a script to install all dependencies using
brew,apt-get,yum,pip,Ret c. - Try to avoid installing dependencies by hand. Use packaged dependencies whenever possible
- Right after installing those dependencies use the same tools to report the installed versions, and save that in a file
- Commit all of this to version control
- Use TSV/CSV to store all types of data whenever possible
- Commit reasonably sized data to version control
- Commit SHA256 of large data to version control
- When the analysis is complete and the project is wrapped up, calculate the SHA256 of all data files, intermediate files, and final products of the analyses and commit that to version control
- Use a workflow manager to run the analysis. I like Make. Many alternatives exist
BrunoGrandePhD
Description
Do you like the idea of reproducible science, but don't know where to start? Have you been told that you should use Git, but never got around to learning it? Join this workshop to learn about various techniques (e.g. R projects, Git and R Markdown) that will help your science be more reproducible. We will start a simple data analysis project from scratch and build it up while following best practices for reproducibility.
Time and Place
Where: Simon Fraser University, Burnaby Campus, Library Research Commons When:
Monday, August 15th, 10:30-12:30 am/ Delayed until fall⟶ Note: This is a two-hour workshop.
Registration
REGISTER HERE
Call for Tips/Links
They say the best way to learn something is to teach it. I'll be doing my research, but I still welcome any feedback I can get on the topic. Feel free to comment below with tips or links on best practices for reproducible data analyses. Thanks in advance!