FOSS Fall 2021 - Week 7, Wednesday

---

Topic: Interactive analyses

FOSS Materials/Useful Links

• Instant Feedback (please complete before you leave class)
• Course Homepage
• Course Schedule
• Today's content

Links

• Reproducible Research in Computational Science
• Ten Simple Rules for Reproducible Computational Research
• Scientific progress despite irreproducibility: A seeming paradox
  
• Project Jupyter
• VICE guide
  • Jason's VICE Webinar
• Create github access token
• Conda package manager

---

Discussion and notes

1. Reproducibility discussion.

2. Use case example: Building a Github to Jupyter workflow using CyVerse.

   Steps

   1. Clone a repository of Jupyter notebooks from GitHub
      to a space on the CyVerse Data Store

      • Demonstrate using mybinder.org
      • Demonstrate using VICE
        • Fork to a personal GH repository
          • git clone repository_url
          • We will try: https://github.com/genomics-education-alliance/leptin-mouse-rnaseq-docker
          • git clone https://github.com/JasonJWilliamsNY/leptin-mouse-rnaseq-docker.git
            • Be sure you use your own username for the clone if you would like to version control this repo ex. git clone https://github.com/UserName/leptin-mouse-rnaseq-docker.git
          • iinit # setup iCommands
          • iput -rP # put copy of your cloned repo in the Data Store
        • Clone using webshell

   2. Launch and edit notebooks using a VICE instance

      • VICE jupyter data science app: https://de.cyverse.org/apps/de/07a2d5b2-76e2-11eb-be5f-008cfa5ae621/launch?saved-launch-id=60054c75-0e80-4169-8a9b-51cba04f756d

   3. Push notebooks back to your forked repository

      • Create a GH access token - "https://github.com/settings/tokens/new # use "cyverse_vice for Note; 7 days experationScopes: repo, write:packages, admin:org, admin:public_key, admin:repo_hook"
      • Configure git on the webshell instance
      • Install Github CLI
        • See: https://github.com/cli/cli#installation
        • conda install gh --channel conda-forge
        • Cash credentials
        • See: https://docs.github.com/en/get-started/getting-started-with-git/caching-your-github-credentials-in-git
        • gh auth login
      • Push back to Github

---

Breakout Discussion questions & notes

1. How do you define reproducible science?

   After discussing on your own, review definitions: Reproducibility vs. Replicability: A Brief History of a Confused Terminology

   • Reproducibility: anyone can understand and replicate the process of the analysis
   • Transparency in your methods, code, data
   • Providing clear steps and tools to reproduce your work.
   • Reproducibility to me means creating a structure and methodology for a project to be easily retested and validated
   • Also add the data used in the paper
   • I can use your data and methods and get the same results
   • Documentation to reproduce the same analysis to get the same results
   • Replicability: reproducibility but with the exact same research question

2. How do these definitions apply to your research/teaching?

Homework Reminders

Homework link will be sent out by Friday!

---