Hello everybody,

I'm planning to submit a package to Bioconductor soon but am having trouble meeting the guidelines. I've got two questions.

1) The functions provided in the package (cerebroApp) follow the analysis of scRNA-seq with Seurat. The input of those functions is the Seurat object itself. One of the requirements of Bioconductor is that at least 80% of the functions contained in the submitted package must have executable example code. My initial idea was to simply contain a very small Seurat object (few cells and few genes) in the package and load that in the example code. However, this increases the package size just enough so that it makes me run into another requirement which is that the package cannot be larger than 5 MB. Does anybody have an idea how to solve this?

2) In the DESCRIPTION file I already have the biocViews: lines, however, the package doesn't depend on any Bioconductor packages so there are no packages listed. During BiocCheck, it complains that ERROR: Add biocViews other than. When I remove that entry entirely, I get ERROR: No biocViews terms found. instead.

I'd be very happy if somebody could give me some advice.

Thank you!

1. If the package is very close to the 5 MB your reviewer might make an exception if there is no way to make the data smaller. They might also suggest submitting the data as an accompanied Data Experiment Package (preferably utilizing the ExperimentHub) . You might also look at the number of experiment data sets already in Bioconductor to see if there is example data/raw files that could already be used. There are a number of available scRNA data available in the ExperimentHub as well.

   library(ExperimentHub)
   hub = ExperimentHub()
   query(hub , "scRNA")
   

   There is also a Bioconductor community slack. I would suggest maybe asking if anyone knows or has provided such data in the #singlecellexperiment channel.

2. biocViews is different than package dependencies. biocViews allow the package to be listed in different sections when listed here. On the left you will notice a drop down menu box of different terms, these are the biocViews . At minimum it seems you would want Technology and singleCell and perhaps a term related to RNA.

3. The fact that the package doesn't depend/import/suggest any Bioconductor package is a "red flag" for us. Your package should rely on some Bioconductor packages. It is generally a requirement to utilize some Bioconductor common data structure at some point in your package; as well as possibly showing how your package will be integrated or utilized in conjunction with other Bioconductor related packages. Consider glancing at the Common Bioconductor methods and classes. Maybe it would be worth while and useful for the community to see if your package could include some conversion of the Seurat object to a Bioconductor object (if they wish but use the Seurat object through out your code base) ? I think maybe monocle package and maybe CellMixS might have done this at some point too with the SingleCellExperiment object? Although I'm not a familiar with the data structures. Exceptions to this have been made if there is a strong argument to justify.

Hope this helps.