developing r&r compendium
josephholler opened this issue · 12 comments
I started mocking up directory structures with informational readme.md
files in each under the Research-Repository
directory.
It doesn't have the json files, makefiles and whatnot of the data carpentry mockup you had found, and modifies the repositories slightly to make more sense. e.g. figures not generated by analysis are included in doc
and there's a structure and .gitignore file for handling private data situations.
the original example (https://github.com/Reproducible-Science-Curriculum/rr-init) used a .json file for metadata. This is hard to read though, unless you're a computer program. I'm wondering if a .csv table wouldn't make a lot more sense, because these are rendered as tables by GitHub.
I created a mock-up of this at school in the COVID-Y1 project folder, but it looks I forgot to push the folder structure up. Stupid. Anyway, you can see my detailed sub-folder structure there, along with some practical tips on use pulled form a couple of sources.
cool. somewhat of a duplication of efforts there, but helpful to also think though it individually.
I like the look of a .csv option for metadata: https://github.com/HEGSRR/RR-Templates/blob/main/Research-Repository/data/metadata.csv
I'll be interested to try fitting the accessibility study into this framework!
No worries. I finished porting it over and integrated with your work. We can discuss Thursday and see how we might want to adjust it.
that was fast! and that's all for now... 12:40am here
Take a look now. Need to plug the preregistration template back in. Last step will be editing the root README.md with index & instructions for the whole repository. I think a 'protocol' folder is sufficient catch-all for any non-code procedures, and I like an index at the root of both folders so that the sequencing of mixed methods research can be clear. Could go back to .md files for the indicies, but I like the ease of editing and viewing .csv files personally, over tables in .md, and they both render smoothly. on GitHub.
I completed a pass through the whole RR Template. If you take a look, we can chat about it an close it this week. I'll deploy it next week for the 4 reproductions in my class.
Sweet. Tues/Wed I'll have a look. I had a short to-do list I was thinking of for this (without seeing if you added these in)
- add metadata subfolder to the data/raw folder, for storing any metadata documentation / print screens from websites where downloaded, etc.
- add processing environment folder under the processing section, for storing metadata about the O.S. / hardware / software environment.
- do you have any thoughts about "results" that are not figures? e.g. tables & outputs of a statistical tests, or "results" that would, in the past, have been included as supplementary data files, but now, the whole repository is the supplement? e.g. video clip, animated data viz, etc. things outside of the framework of in-line paper content.
Addressed 1 and 3. Not sure where you want 2, so I'll leave it to you.
Ok! note to myself: check the .gitignore
one last time to include any raw metadata
Let's plan to discuss systematic revisions to the template in mid-May. My class will finish their 4 reproduction in about 2 weeks and I will go through a round of editing and harmonizing their reports for the subsequent paper. I will track those in this repository. If you do the same, we can rectify them in mid-May and parcel out who will change what.