In ropenscilabs/taxa: Classes for Storing and Manipulating Taxonomic Data

options(max.print = 100, width = 100)

# devtools::install_github("ropensci/taxa", ref = "vectorize")

library(taxa)

I used the vctrs package to try out making S3 vector-like versions of classes to hold taxon databases and taxon id. If this works out, I would like to do the same for taxon names, ranks, and taxa as a whole, which would include the ids, names, and ranks. The ids, names, and ranks each can hold their own database ID.

taxon_db

The taxon_db class is just a character vector of database names with extra validity checks to make sure that names are valid.

taxon_db()
taxon_db('ncbi')
taxon_db(rep('itis', 100))
taxon_db(rep('itis', 2000))

The acceptable databases are stored in database_definitions in the same way knitr chunk options are stored and changed.

database_definitions$get()

All names in taxon_db objects must be in this list or errors are thrown:

taxon_db(c('ncbi', 'itis', 'custom', 'tps'))

This also happens on assigning values:

x = taxon_db(rep('itis', 10))
x
x[2:3] <- "ncbi"
x
x[2:3] <- "custom"

However, you can add new databases to the list:

database_definitions$set(
  name = "custom",
  url = "http://www.my_tax_database.com",
  description = "I just made this up",
  id_regex = ".*"
)

And then you can use that name too:

x[2:3] <- "custom"
x

taxon_id

The taxon_id class stores taxon IDs and their associated database (as a taxon_db vector).

taxon_id()
taxon_id(1:10)
taxon_id(1:10, db = 'ncbi')
taxon_id(1:10000, db = 'itis')

Run code in an R console to see colored output. I forgot how to make it work in Rmarkdown.

Since a taxon_db object is used, database names must be valid:

taxon_id(1:10, db = 'custom2')

The database values can be accessed and set with taxon_db:

x <- taxon_id(1:10, db = 'ncbi')
x
taxon_db(x)
taxon_db(x) <- 'itis'
x
taxon_db(x)[1:3] <- 'tps'
x

... if they are valid:

taxon_db(x)[1:3] <- 'custom2'

If a database is defined, then the ids must match the ID regex for that database in database_definitions$get():

taxon_id(letters[1:3], c('ncbi', 'itis', 'itis'))
taxon_id(letters[1:3])

You can set taxon id values with numbers or characters, but only taxon_id inputs can supply a defined database.

x <- taxon_id(1:10, db = 'ncbi')
x
x[3:4] <- 1
x[6:7] <- c("a", "b")
x[8:9] <- taxon_id(1, db = c('itis', 'tps'))
x

Use in tables

The taxon_id prints well in data.frames and tibbles, using color when possible:

data.frame(x = taxon_id(1:10, "ncbi"), y = taxon_db(rep("itis", 10)), z = 1:10)
tibble::tibble(x = taxon_id(1:10, "ncbi"), y = taxon_db(rep("itis", 10)), z = 1:10)
str(taxon_id(1:10, 'ncbi'))

Plans for taxon name and rank

taxon_name and taxon_rank would be very similar to taxon_id. taxon_rank would either be based on a factor, or be factor-like and would have database-based validity checks for rank names and order.

Plans for taxon

taxon would contain at least taxon_name, taxon_rank and taxon_id vectors. It would probably also have authority information. It could also have field for arbitrary information, but this might not be needed since it could be in a table that could contain that info in other columns.