Recommended Knitr settings for R Markdown docs?

[kubu4]

Anyone willing to share some examples of their Knitr settings they like to use? Especially for R Markdown docs with heavy usage of Bash chunks (not sure if this impacts setting choices or not)?

[sr320]

Here is an example
https://github.com/course-fish546-2023/steven-coursework/blob/main/assignments/code/3.1-blast.Rmd

the key is for bash code your chunk should an r chunk, indicating bash engine - example

```{r blast-look, engine='bash', eval=TRUE}
head -2 ../output/Ab_4-uniprot_blastx.tab
```

[kubu4]

To clarify this, do you replace eval=TRUE with cache=TRUE?

Basically, I'm having difficulty getting a compute-intense chunk to evaluate (basically, I'll like the run time to be displayed) and have it cached, so I don't have to re-run it over and over when I tweak other chunks or add additional markdown to the other portions of the document.

[sr320]

do you replace eval=TRUE with cache=TRUE?

Yes - that is correct. Note the first time you knit it will run, but after that it will used cache.

Just curious- why run at all in knit? as opposed to just running chunk prior to knit.

[kubu4]

If I don't eval=TRUE while knitting, then the output from the chunk isn't displayed in the knitted output. Isn't that correct? I guess I'm confused on what's happening when knitting...

[sr320]

Correct, the chunk will not run when you knit. This would be key if generating a plot.

If you were just creating files in chunk, you could run a following chunk to display (eg, cat, head) which likely would take less time. Example is when I run HiSat, stringtie, convert SAMs, I do not eval but eval following chunk eg -https://github.com/urol-e5/deep-dive/blob/main/D-Apul/code/05.32-lncRNA-discovery-overview.md#2-stringtie

[kubu4]

Arlgihty, thanks. I think that helps me understand how it all works.

Maybe I just need to take a different approach to the long-running chunks (e.g. don't have additional code in those chunks that prints stuff).