In the move towards open science, it is becoming common practice to deposit the data accompanying a publication in a dedicated data repository.
Often, authors prefer to delay the publication of the data set until after review of the manuscript is completed. To facilitate this, many data repositories provide mechanisms to (1) reserve a DOI for the data set before publication, and to (2) generate a private link to the data set for the review phase.
My question is: As an author, what are best practises for citing the (unpublished) data set in a submitted manuscript?
One solution is to put the dataset DOI in the submitted manuscript (although it does not resolve yet), and to communicate the private link through some separate channel that the journal may provide during submission (but that the reviewer may not read?). Another solution is to put the private link in the manuscript, meaning one needs to remember to replace it with the DOI during the proof stage.
Neither solution seems ideal and I'd be interested to learn whether better ones have been figured out.