Internal helper for markerParentageLikelihood (issue #155,
docs/planning/issue155-parentage-likelihood-candidate-lookup-plan.md
D1). getPotentialParents only ever builds a candidate list
for an animal with at least one missing (NA) parent slot; a flagged
animal whose recorded parent is present but Mendelian-inconsistent has
both slots non-NA by definition, so it never appears in
getPotentialParents's own candidate search. This helper
returns a local, unmutated-in-place copy of pedigree with exactly
the named (id, role) slot(s) set to NA, so pedigree merely
looks like a genuinely-missing-parent case to
getPotentialParents -- which is called completely
unmodified against it, reusing its full
demographic-eligibility engine (breeding-age floor, gestation window,
proven-breeder preference) at 100\
never changed -- the copy is scoped only to the caller's own
getPotentialParents call argument.
.markerFlaggedSlotPedigree(pedigree, ids, roles)A copy of pedigree, un-mutated except for the named slot(s).
A duplicated pedigree$id is left un-blanked -- which row is "the"
flagged animal is undefined for an ambiguous id, so this falls through to
the same fail-soft, zero-candidates posture as an id genuinely absent from
pedigree, rather than silently blanking every matching row (mirrors
scoreOnePair()'s own defensive nrow(pedRow) == 1L pattern a
few lines above in this file).
a data frame with (at least) columns id,
sire, and dam.
character vector of flagged ids, one entry per (id, role) pair.
character vector of flagged roles ("sire"/"dam"),
the same length as ids, paired positionally.