search

thelovelab / tximeta

Transcript quantification import with automatic metadata detection
https://thelovelab.github.io/tximeta/
67 stars 11 forks source link

Generate Methods section boilerplate text #23

Open mikelove opened 4 years ago

federicomarini commented 4 years ago

Just chiming in, since I have recently always tried to have some kind of functionality like this in my recent packages. If you want to discuss some aspects, happy to exchange some opinions 😃

mikelove commented 4 years ago

Sure, so I was thinking of this because of some discussion from the CWL folks about generating Methods section text with versions.

Given that tximeta knows the Salmon / alevin version and the source and release for the transcriptome for human, mouse, and drosophila, it's just a paste command to generate a bit of text like:

Salmon version X [cite1] was used to quantify expression of the GENCODE [cite2] release X human reference transcripts.

[cite1] Patro et al 2017 ... [cite2] Frankish et al 2018 ...

Tximeta could also add other details like:

The average number of mapped reads across samples was XX.X million [min, max].

Can you point me to what you've added to recent packages?

federicomarini commented 4 years ago

Ok, I thought you wanted something more extended.

What I was referring to is for example what I did in pcaExplorer, ideal, and more recently in GeneTonic. These are all Shiny apps where I also include an Rmd document with the app that picks the values from reactive elements (or for GeneTonic some parameters passed "offline", as in batch mode), and compile the HTML document. See here e.g. for the report of GeneTonic (called via happy_hour, because naming matters indeed): https://github.com/federicomarini/GeneTonic/blob/master/inst/extdata/cocktail_recipe.Rmd

mikelove commented 4 years ago

Ah cool!

Yeah, I guess the title of this feature request is overly ambitious. I just mean Methods section boilerplate text. I should clarify.

federicomarini commented 4 years ago

Well it would still make the feature highly desired. It's not that we are lazy (or in other terms, we tend to optimize the usage of time 😄 ), but often times we do errors in doing things manually, and exactly in these cases a machine-automated solution is just the way to go