I have been using GitHub to collaborate/share data analysis (and a bit of writeup) with my co-authors/collaborators for the last year or so. As most journals now require (or at the very least recommend) submitting original data used in the analysis, alongside the relevant scripts/codes, during the manuscript submission process, I thought it was easiest simply to submit the persistent link to the GitHub repository containing all the data and the scripts/codes used in the analysis and visualisation as part of the manuscript submission.
While I kept the repository private during early phases of data analysis and writing-up, once we were ready to submit the manuscript, I made the repo public so that journal editors and reviewers can look at our data and codes, and replicate the entire analysis if they wanted to. This not only makes the whole process transparent, it also helps us get feedback on this part of our research directly if those reviewing the manuscript wish to do so.
The paper is currently under review, and the repository can be accessed here.
Update: The paper has since been published in PeerJ journal and can be accessed here.