Presentation outline

[cmsmith91]

1. Intro - Colleen

• What is reproducibility?
• Why aim for reproducibility?
• 4 facets of reproducibity (u of uthrecht)
• Fig of breaking code: https://learning.nceas.ucsb.edu/2023-04-coreR/session_18.html
• What will give you a good return in investment? One min in data management saves you 15 min of headache down the line

2. Version control - Camila

• Why important?
• What is it?
• Example

3. Organization - Camila

• R projects -each project its own directory and gives relatives paths (portable code)
• Different files: raw_data (don’t touch), figures, processed_data
• File names – order, [inputs/outputs match]
• Camila uses Google Drive & EDI (like Zenodo)
• where to put raw data?

4. Documentation - Colleen

• Read me file
• Writing reproducible code
• Stick to a readable coding style (styleR package)
• Naming variables and functions
• Commenting code: why instead of what
• Functions
• Defensive programing: errors[CS3]
• Github issues – good for communicating with collaborators,
• Github wikis – particularly for documentation; markdown file that you access through github

5. Automatization & Dissemination

• What to do about dependencies? (sessionInfo, packrat package)

other potentially useful resource: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1005510

[camilavargasp]

Finally, putting some thought into how to streamline all the great content we have put together. I would like to slightly restructure things to make sure we have a good flow and transition. Below I present the new outline indicating what is ready and where we still need to add some content.

Introduction

• [x] What is reproducibility?
• [x] Why aim for reproducibility?
• [ ] Tools that help make our work "more" reproducible and transparent.
  • [ ] Programming language
  • [ ] Version control
  • [ ] Authoring systems
• [x] But, making your work truly reproducible is hard --> Example from Collen's work
• [x] A good starting point: Aim to "do our data exploration, analysis final figures or tables prep and write text in a cohesive workspace so that our publications are automatically updated when out analysis is updated. (Streamline the process from raw data to final report, keeping track of the changes and decisions, avoiding copy/pasting)
• [x] Today we will talk about 5 aspects to achieve scientific reproducibility: Organization, Documentation, Setting up a workflow, Automation, and Dissemination. (Note: bulk will be on the 3 first ones)
• [x] Disclosure: All our examples are using R as a programming language and RStudio as the IDE

Organization Setting a base structure from where I am going to work?

• [x] Where am I doing my computational work?
• [x] Where do files live?
• [x] How do I tell my code where to find things?
• [x] File Paths
• [x] Example with R Projects
  • [x] Organizing things within a project
  • [x] Here function in R
• [ ] How do I keep track of my work?
  • [ ] Analysis is an iterative process (files change through time)
  • [ ] Data Management best practices (Keep raw data raw, scripted analysis, keep track of changes)
  • [x] Version control (What and Why, Tools: Git and GitHub)
• [ ] Transition: Good organization is the first step to good documentation

Documentation

• [x] Why is it so important
• [x] How to document a project
  • [x] README: Tell others about your project, describe how your project is organized, what are the steps needed to get to the end result.
  • [x] Naming conventions
  • [x] Code Style (Code readability, naming variables and functions, commenting code, functional programming, literate analysis (need to add))
  • [ ] Communication with collaborators and documenting along the way (Tools: Issues and Wikis)

How does this all work together? [Need to figure out a good name for this section]

• [ ] Review all the tools (R, Git, GitHub)
• [ ] Where do we keep our big files?
• [ ] 3 examples: Google Drive, Zenodo, Server.

Automation and computational reproducibility

• [ ] targets package
• [ ] preserve your computational environment
• [ ] Research compendium example using rrtools?

Dissemination [Or publishing and sharing]

• [ ] Something brief along the lines of this
• [ ] And this