andrew-edwards / ednacutoff Goto Github PK

@dcallander @bensutherland

You should be able to install the package and run the vignette. See the README file here for instructions.....

https://github.com/andrew-edwards/eDNAcutoff

This repo is public but shouldn't make much sense to anyone without the manuscript (which is private) - having it public does make testing easier.

Okay, let me know if anything works! The structure is all set up nicely so it should be relatively easy to add new methods etc., in conjunction with updating the manuscript.

If the vignette doesn't work then you probably won't get too far - just let me know (here).

Maybe try this (and put a short example in the readme):

What about My Vignettes?
So how should you make your vignettes available to the public? What I’ve done is commit both the .rmd and generated .html file to my GitHub repository. GitHub won’t directly render the .html file, but you can use the GitHub HTML Preview service. Basically, you just give it the url of your html file from your GitHub and it will render it. For instance, here is the preview of my html vignette:

http://htmlpreview.github.io/?https://github.com/tinyheero/CHL26predictor/blob/master/vignettes/introduction.html

• Do simple example as a unit test
• Do complex example (anonymise the species) as a unit test
• Write a vignette
• Add test for change_alpha
• Change terminology (mock etc.) as per andrew-edwards/eDNA-filter#17
• Add README.Rmd to make README.md
• Get Travis running to check commits don't break anything
• Do a check() to check submittable to CRAN down the road
• Get someone to test a download

Clean R session, this works:

load_all()
build_vignettes()

but

install(build_vignettes = TRUE)

fails and can't find small_example.

• Add any table numbers to dataset descriptions in R/data.R

A while back https://github.com/andrew-edwards/eDNA-filter/commit/733b7b7d0cf13fa51c351dd6bb29f6a2db7f4813 Davon confirmed that my result from the function matched hers. Need to double check that again (as it's not clear when or what I gave her), and then it will be automatic in the unit test.

Can't just do

> small_removed <- remove_false_pos(small_example)
> small_removed - small_example

because the first column contains characters:

> small_removed
# A tibble: 4 x 8
  Sample     m1     m2    m3    m4   `1`   `2`   `3`
  <chr>   <int>  <int> <int> <int> <int> <int> <int>
1 mock   257329 169668 79560 47859     0    17     0
2 1           0      0     0     0    42    17 61802
3 2           0      0     0     0 43997    10    17
4 3           0      0     0     0    26     0 22621

So function just needs to remove first column, do the difference, then put the first column back in again. Presumably something like

res <- dplyr::select(input, -1) - dplyr::select(output, -1)
res <- cbind(dplyr::select(input, 1), res)      # or dplyr version of cbind