prettifying vignette revision (formatting, hyperlinks etc.)

[HelenaLC]

(Note: branched out from solve-conflict-tidyverse PR #69, hence these changes are included here as well. Alternatively, could do a separate PR once #69 has been resolved.)

Vignette revision as discussed at stemangiola/tidyseurat#66, including:

• loading full set dependencies in the preamble
• fixing vignette HTML rendering on Bioc (see here)
• hyperlinking external packages using BiocStyle::CRAN/Biocpkg() and friends
• consistent style to improve readability and distinguish between, e.g.,
  ecosystems (tidyverse), classes/variables (tibble), functions (unnest())
• prettify table summarizing tidySingleCellExperiment's functionality

...perhaps best to look at the HTML version of this, and compare to the Bioc one for review!

[stemangiola]

Looks great. However, branching from #69 is not great because it fails the checks for a reason I could not solve so far (cannot find filter method due to some roxygen mystery).

Apologies, would you mind branching from master in this case and putting your new vignette there?

Thanks!

[HelenaLC]

Sure, no problem, will do! Re documentation: I have a running version that is warning-free (at least for me, I was getting devtools::document() warnings previously). That PR will look huge (1000s of character changes), as I am dropping all manual documentation (except for utility functions), and am inheriting most from dplyr/tidyr/… That been said, I propose a PR with 1 dplyr and 1 tidyr function; check those man pages for whether they’re acceptable; then PR the rest. Or you have a suggestion?

I think it’ll be super nice in that long-term maintenance will be much easier, and it’s as comprehensive as we can be. Plus, bringing script lengths down by 100s of characters makes it a lot neater. And we can focus future efforts on useful (transcriptomics) examples.

[stemangiola]

I am dropping all manual documentation (except for utility functions), and am inheriting most from dplyr/tidyr/

sounds good, the only specific thing are examples, which use SCE instead of a tibble, and little more. So more we manage to use from tidyverse and better it is.

That been said, I propose a PR with 1 dplyr and 1 tidyr function; check those man pages for whether they’re acceptable; then PR the rest. Or you have a suggestion?

I did not understand this one.

[stemangiola]

maybe

• one PR from master with vignette and
• one PR from master for documentation

But if you have done all together it's also fine. Happy to review it when it's ready.

[HelenaLC]

All I meant was that I could do a small PR with 1 function documentation update, so it’s easy to check the “new style”, or do all at once (but that’ll affect maaany files and might be difficult to review)- you’re call 🙏