Our top priority is doing reproducible science. This means establishing an efficient workflow that allows for collaboration in a convenient manner. The following set of guidelines and suggestions aim to this objective.

Onboarding

There are a few things that each new member should do when he/she joins the research team. These are:

• Create a GitHub account.
• Check that your owncloud account is working.
• We use the slack for communication and wrike for project management. Make sure that you have been invited to both of them.
• Download and begin to maintain a reference manager. We recommend Zotero, as the free software interfaces well with both Microsoft Word and Google Docs.
• Read the common scripting practices described here.

Data management

• We generally use a folder called "data" within each repository.
• We keep our raw data in the github repository related to the project, unless the data files are too large.
• We store our raw data with metadata describing what’s in the file and what the columns mean. We consider these data as read-only.
• If we clean the data, we often use a folder called something like "raw" to differentiate data in its original form from data that has been manipulated.
• If we are using data downloaded from another data source, we include the data source in a README.
• If our data are too large to store on GitHub (file size> 100 MB), we store them in owncloud and include the link to README file for reproducibility.
• Some useful suggestions and ideas can be found in the Cambridge Data Management Website.

Analysis

• We do our data analysis in GitHub repositories to facilitate collaboration and sharing.
• We use scripts to process data, make models, do analyses, etc. and avoid spreedsheets, gis applications or other software.
• R is used by the most team members and the data.table, ggplot2 packages in specific. Of course other software/packages are welcome.
• We try to comment a lot in our code.

Results-Presentation

• We aim to fully reproducible papers, as in this example.
• We publish git repositories through Zenodo upon publication of a manuscript.
• Collaborative manuscripts are written either in Overleaf or Google Drive.

Guides and tutorials

• R for data science
• R/R studio for pure beginners
• Rstudio with GitHub
• R with git
• Zotero
• data.table vignette
• writing good software

Acknowledgements

Much of the inspiration for preserving an efficient, reproducible workflow came from openscapes, while some of the material used here were copied from Pinsky lab.