/project_template

Use this project template to organize your projects. Inspired by work of many others. Check:

To start using this template download the repository from here, change repository name and start doing great science

Some tips and recommendations.

data

  • In data you should store only raw data.
  • This data will be read when you do your analysis but can never be modified.
  • If you want to create modified data-sets from this data save it in the output folder.
  • Create a document where you explain how you obtained the data and also the meaning of all your variables.

docs

Here is where your article/thesis goes. You can use Latex, or Markdown. If your article is heavy on R analysis you could use Rmarkdown to directly write your article then knitcitations is highly recommended to handle citations.
My personal workflow involves using Atom locally, write directly in Github´s editor or using Prose.io. I use magic citations (that work on both Atom and online with Chrome) from Papers to insert citations. I will be happy to buy you a student Papers license if you want to go that route. To generate References and export to Latex/Pdf via pandoc I use a small makefile script. Please check Karl Broman´s tutorial on make. I have included a sample in the docs folder. With this strategy you:

  • Write in Markdown (you can work locally or online) which I think is easier for beginners.
  • Insert citations with Magic Citations from Papers. But you can use Mendeley or other solutions. Is easy to generate a bibtex bibliography file in Papers or Mendeley and then copy it in a bibliography.bib file.
  • Keep your article version controlled by committing frequently to Github.
  • Generate PDF or latex file using a simple makefile.

figures

Save here your final figures. The ones that you will use to generate article PDF.

output

  • Output is for data you generate with your scripts, figures etc. The idea is that everything that you put here is disposable because it can be regenerated via scripts.
  • You could subdivide this folder in two: figures and data.

scripts

Scripts of your analysis. If in R, Rmarkdown is highly recommended. If the project is simple you could have a single Rmarkdown file stored in the root of the repository and skip this part. An Rproject file is also generated in the root of the repository

Internal communication

Finally, for each project we will generate an associated gitter channel as a way to keep track of ideas, suggestions etc.