Collect every cross-center identity-mapping problem, without stopping

The "show every problem at once" companion to resolveCrossCenterIds (issue #149 Slice 1), sharing its five validation checks – id existence, mapping uniqueness, undeclared id collisions, a pedB animal that reuses an id the mapping gives a pedA animal, and conflicting recorded parents – via the same internal helpers (D2, docs/planning/issue149-cross-center-identity-mapping-workflow-plan.md section 3). resolveCrossCenterIds() stop()s on the first problem it finds; checkCrossCenterMapping() never stop()s on a domain problem – every one found becomes a row in the returned data.frame instead, so a curator can see and fix every issue at once rather than one at a time. A structural problem (a required column missing from any of the three inputs) still stop()s immediately, matching every other checkXxx() function in this package (checkKinshipOverrides, checkTwinRelations).

Usage

checkCrossCenterMapping(pedA, pedB, mapping)

Arguments

pedA

a pedigree data.frame for the first center, with (at least) columns id, sire, and dam.

pedB

a pedigree data.frame for the second center, with (at least) columns id, sire, and dam.

mapping

a data.frame with columns idA and idB: one row per curator-proposed cross-center identity link.

Value

A data.frame of every domain problem found, with columns type ("existence", "uniqueness", "collision", "mappedIdClash", or "conflict"), ids (the offending id(s), as a single comma-separated string), and message (a human-readable description). Zero rows means none of these five kinds of problem was found. A "mappedIdClash" row names an id of a pedB animal with no mapping row that the mapping already uses for a pedA animal; a merge would hold two rows with that id. resolveCrossCenterIds can still stop on conflicting values in other columns shared by both pedigrees (e.g. sex), which this function does not check.

Details

Existence and uniqueness problems (tier A) are checked first; if either is present, only those are returned and the collision, reused-id and conflict checks (tier B) are skipped entirely, since a mapped id that does not resolve to a real pedigree row makes those checks meaningless. Tier B – and all three of its checks, across every mapped pair – runs only once tier A is clean.

See also

resolveCrossCenterIds

Examples

library(nprcgenekeepr)
pedA <- data.frame(
  id = c("P1", "P2", "T1"), sire = c(NA, NA, "P1"), dam = c(NA, NA, "P2"),
  stringsAsFactors = FALSE
)
pedB <- data.frame(
  id = c("X9", "O1"), sire = c(NA, "X9"), dam = c(NA, NA),
  stringsAsFactors = FALSE
)
mapping <- data.frame(idA = "T1", idB = "X9", stringsAsFactors = FALSE)
checkCrossCenterMapping(pedA, pedB, mapping) # zero rows: clean
#> [1] type    ids     message
#> <0 rows> (or 0-length row.names)