Migration to Quarto

[lgeistlinger]

Hi @PeteHaitch @alanocallaghan:

I wondered whether you had any thoughts re: migrating the book from using rmarkdown over to using quarto? While this might be a bit of a hassle in the beginning, it will give a set of additional features and layout options, assuming also that quarto will supersede rmarkdown in the long run.

It seems some other folks have recently successfully migrated their books eg. the Modern Statistics for Modern Biology (MSMB) book by Holmes and Huber. Also @js2264 has put together useful functionality for this purpose in his BiocBook that would likely help with that migration.

@andrewGhazi, a data scientist here with us at CCB, actually took a first shot at this here: https://github.com/ccb-hms/OSCA.intro.qmd and deployed the rendered version: https://ccb.connect.hms.harvard.edu/osca_intro_qmd/.
That seemed pretty straightforward, subtracting a couple of issues that he describes in the README of the repo.

@js2264: as you seem to have thought extensively about these book setups, do you see any major problems with migrating the current rebook-based setup of the OSCA book over to your BiocBook framework? From this comparison, it seems that we would only need to put some more thoughts into reuse of objects between sub-books and cross-references between sub-books?

Also tagging @LTLA @hpages @vjcitn in case you have any thoughts on this.

[hpages]

No particular thoughts on this but it would be interesting to know what quarto features and/or layout options exactly you're interested in.

FWIW I don't think quarto will supersede rmarkdown in the long run. Nobody should worry about that 😉 The former being a superset of the latter IMO it only makes sense to use the former if you're actually going to use the additional features.

[PeteHaitch]

I'd also be interested to know what quarto gets us that we don't have (and need).
I'm very reluctant to change without compelling reasons.

[alanocallaghan]

I'd be fine with testing quarto for new material but not with migrating existing stuff to quarto unless there's a strong reason to switch

[Varix]

agreed with @hpages , I can understand the idea of potential future proofing for a future feature if you were starting a new project from scratch right now. but spending the time to convert now, for the possibility of new future unsupported in rmarkdown feature we want may not be the best ROI on time.

[lgeistlinger]

Thanks all. That’s helpful. I agree that we want to be conservative with (maybe not ultimately necessary) changes to the currently stable build of the book.

When it comes to reasons for using quarto over rmarkdown, I guess those are similar reasons that made users choosing rmarkdown over sweave and ggplot2 over base R plot - it’s a somewhat modernized approach towards the same end.

Other benefits seem to include:

1. fewer dependencies - currently we need rmarkdown, bookdown, and rebook to build the book, that would all be handled by quarto
2. R and Python code blocks in the same document - that is straightforward with quarto and will be helpful for the new R/Python sub-book
3. (Subjectively) nicer looking outputs

I am with Alan though - let’s leave the old stuff as it is for now and experiment some more with quarto for the new stuff. And then compare the products and reconsider.

[alanocallaghan]

R and Python code blocks in the same document - that is straightforward with quarto and will be helpful for the new R/Python sub-book

https://bookdown.org/yihui/rmarkdown-cookbook/eng-python.html