Check names in Breedbase before building a design — breedbase

Run this first, before designing the trial: it checks that the location(s) and every entry and check are already in Breedbase (and, when given, the breeding program, the folder and that the trial name is free). All problems are reported together, with close matches as suggestions (for example "ibadan" -> "Ibadan"), so they can be fixed in one go. save_breedbase() repeats the same checks before writing.

Usage

breedbase_check(
  con,
  accessions = NULL,
  location = NULL,
  program = NULL,
  folder = NULL,
  trial_name = NULL
)

Arguments

con: A connection from breedbase_login().
accessions: Entry and check names: a character vector, or a design from augmented_row_column_design(), augmented_prep_design() or resolvable_row_column_design().
location: Breedbase location name(s).
program, folder: Optional breeding program and folder (the folder is optional for saving too: without it the trial goes directly in the program).
trial_name: Optional trial name. A trial with this name that already exists (with plots) is reported to fix.

Value

Invisibly, a tibble with one row per name checked: what, name, status ("OK", "MISSING", "NEW", "TAKEN", "EMPTY" or "REPEATED") and suggestion. Its attribute "ok" is TRUE when nothing needs fixing.

Slow servers

Accessions are searched 25 names at a time, because Breedbase looks up the pedigree of each one. If the server times out (HTTP 504, 502, 503 or 429), the batch is split in half and tried again, so a busy server is slower but still works. A line is printed after each batch with the names done and the time. Accessions found are remembered until R is restarted, so save_breedbase() after breedbase_check() does not search them again.

When creating the folder, the trial or its plots times out, the save does not stop: the server may still finish the job, so it checks for the result every 20 seconds for 3 minutes, and tries once more if it is not there (Breedbase refuses names that already exist, so this cannot make duplicates). If a save is cut off anyway, run save_breedbase() again: an existing folder is used, and a trial left without plots gets its plots. Settings: options(breedtools.write_wait = 30, breedtools.write_checks = 10). Settings: options(breedtools.germplasm_batch = 10) for smaller batches, options(breedtools.breedbase_timeout = 600) to wait longer (seconds) for each answer.

Examples

if (FALSE) { # \dontrun{
con <- breedbase_login("https://cassavabase.org", "jdoe")
entries <- c(paste0("IITA-TMS-IBA", 1:108))
checks <- c("TMEB419", "IITA-TMS-IBA30572", "TMEB693", "IITA-TMS-IBA000070")
breedbase_check(con, accessions = c(entries, checks), location = "Ibadan",
                program = "IITA", folder = "AYT 2026", trial_name = "AYT_2026_Ibadan")
} # }