Breakout rooms 1/3 - Visualizing data and Results & Research documentation #

Markdown CheatSheet
Our Github Repo
Google slides

We’ll cover the following topics: #

Step 1: Presentation of the breakout room topic
Step 2: Brief tutorial and time for individual practice
Step 3: Discussion and summary of challenges
Step 4: Return of participants and organizers to the main workshop room

Examples / References that display research documentation, data, and results #

Use of displaying notebooks as experiments on Github

• Team working on optimization of a neural network: https://github.com/DiscoveryDNA/team_neural_network/tree/master/code/experiments
• “Toolchain walkthrough” for organising data for systematic review: https://github.com/softloud/sysrevdata (https://softloud.github.io/sysrevdata/)

Working on a team using Github Organizations

• Overarching research team that works on multiple projects. Seperating projects that range from experiments to software: https://github.com/DiscoveryDNA
• An ugly example of team working: https://github.com/LivingNorway

Hosting website to work on research publicly

• Organization that shares tutorials and research. Each project will evolve into a “post” on website https://github.com/cabinetofcuriosity
• Website: https://github.com/cabinetofcuriosity/cabinetofcuriosity_site

Documentation of data standards

• https://github.com/ess-dive-community

Archival of code and results for papers

• Largely non-computational project. Code was just used to analyze resutls and create graphs: https://github.com/iamciera/sister-of-pin1-material/blob/master/cosegregation/readme.md

• Largely Computational Project: https://github.com/iamciera/lcmProject

Learning Github

• https://github.com/hlowman/TidyTuesday/blob/master/resources/Git_GitHub_Guide.pdf
• https://afredston.github.io/learn-git/learn-git.html

Discussion questions: #

1. What other tools have you used for this purpose?
   • Sharing files back and forth
   • Microsoft Office with track changes
   • OSF R package
2. How does GitHub compare to other tools you’ve used for this purpose?
   • Harder to get some collaborators on board
   • Smaller file size limits
   • Undergrads can be the best adopters, helping with data tidying (no habits), one workshop at the beginning
   • microsoft owns github
   • size limits actually quite large on OSF
3. What are the main challenges in using GitHub for this purpose?
   • Making sure everyone on team is willing to use
   • Size limits
   • Certain file types have cryptic ‘diff’ documentation in Github
   • large corporation owning code
   • terminology not consistent across programs
   • Working with a team with varying skill levels and associated challenges
   • seems like a big step relative to track changes in MS word - some collaborators still reluctant to even use Google drive
4. What are the main benefits of using GitHub over some other tool?
   • Making sure all collaborators are using the same version
   • Example of a microsoft access database- collaborating with several students including undergrads
5. What could be done to improve GitHub for this purpose?
   • Glossary of terms put into user-friendly language
   • teaching git starting with 5 commands, not overwhelming with all vocabulary
   • teasing apart which software engineering principles are needed for researchers, vs. what should be delegated to a software engineering role within a research team
   • Explain to students that it’s very hard to ‘break’ it
   • Starting with low stakes projects
   • Make learning project-based
   • You don’t have to use github if you use git: Gitlab instead of Github, but same fallbacks. Gitlab offered free private repos
   • When teaching Git/Github, it’s important not to transfer hesitancy to students – helpful to model behaviour of not needing to be an expert and making mistakes

Misc discussion #

• Some members seeing big disjunct between their willingness to adopt new technology relative to PIs - writing manuscripts as markdown documents, converting into word, getting back track changes and implementing manually
• github pilot - helps autocomplete code, took everyone’s code and did ML including all license types
• people like software developers experience more top down support/pressure to adopt, whereas researchers usually working from bottom up
• dealing with big files - splitting and binding big files can help
• adding releases helps get around file size restrictions - unlimited file size
• osf R package as an alternative for written components of projects, scripts go onto github
• As a minimum situation where no new software needs to be learned - build timestamp into all documentation (including data download)
• most software engineers don’t store data either locally or online – this is why there is a limit