Parent-offspring conflicts between the genotypes and the pedigree

For every animal whose sire or dam is genotyped together with it, counts the markers called in both and, among them, the opposite homozygotes (0 in one, 2 in the other), which Mendel forbids between a parent and its offspring, genotyping error aside. A rate above exclusion_threshold says the pedigree or the sample is wrong: a parent recorded by mistake, a sample swap. The rule and the default 1% are the exclusion test of preGSf90 (verify_parentage, exclusion_threshold 1; Hayes 2011), with one difference: preGSf90 divides by every SNP, and here the denominator is the markers called in both, so a pair with many missing calls is not diluted. A true pair shows only the genotyping error, and an unrelated pair 2 p^2 q^2 per marker (12.5% at p = 0.5), some 5% to 10% over a common panel. The column of maternal grandsires of a sire_mgs() pedigree is left out, since grandparent and grandchild can be opposite homozygotes.

Usage

parentage_check(
  pedigree,
  genotypes,
  exclusion_threshold = 0.01,
  sex_markers = genotypes[["sex_markers", exact = TRUE]]
)

Arguments

pedigree: data.frame animal, sire, dam (or a sire_mgs() pedigree). A repeated animal is refused; a genotyped animal missing from it is a warning, and none at all an error, because no pair would read as no conflict
genotypes: list with ids and m (0/1/2, NA for missing), and optionally sex_markers
exclusion_threshold: fraction of opposite homozygotes, among the markers called in both, above which the pair is a conflict
sex_markers: the columns of m on the sex chromosomes (indices, names or a logical vector), left out of the test; by default genotypes$sex_markers, and NULL tests every marker

Value

a data.frame with one row per genotyped parent-offspring pair, in pedigree order: id (the offspring), parent, type ("sire" or "dam"), markers (called in both), opposite (opposite homozygotes), rate (their fraction) and conflict (rate > exclusion_threshold)

Details

Use autosomal markers: PLINK codes a male hemizygous X as homozygous, and a son takes his X from the dam, so a sire and son are opposite homozygotes on about 2pq of the X markers with no error at all; with 3% of a chip on the X that alone is about 1%, the threshold. preGSf90 skips the sex chromosomes in this test (sex_chr); here sex_markers = does it, with the map that read_plink() returns, or a genotypes$sex_markers element, which qc_genotypes() sets when it is given them.

Unlike the fitters, this function does not refuse an animal with a low call rate: it imputes nothing, and each pair is counted over the markers called in both.

Every single-step fitter runs the same check, on every marker it is given but those that genotypes$sex_markers names, and warns, naming the pairs, when one exceeds 1%; it then fits the data as given. The two remedies: set the wrong parent unknown in the pedigree (below), or drop the offspring's genotype, which is what preGSf90 does by default and what qc_genotypes(genotypes, pedigree = ) does. options(BreedingR.verify_parentage = FALSE) (or 0, the verify_parentage 0 of preGSf90) turns the fitters' check off, for genotypes simulated without the pedigree, where every pair conflicts by construction; this function and qc_genotypes() still run it.

References

Hayes, B.J. (2011). Technical note: Efficient parentage assignment and pedigree reconstruction with dense single nucleotide polymorphism data. Journal of Dairy Science 94:2114-2117.

Misztal, I. et al. (2018). Manual for BLUPF90 family of programs, preGSf90 (verify_parentage, exclusion_threshold, sex_chr).

Examples

s <- simulate_breeding(20, 2, 20, n_markers = 300, seed = 1)
g <- s$genotypes
ped <- s$pedigree
# a wrong sire for the first genotyped offspring
filho <- which(ped$sire %in% g$ids & ped$id %in% g$ids)[1]
ped$sire[filho] <- setdiff(g$ids, c(ped$id[filho], ped$sire[filho]))[1]
ck <- parentage_check(ped, g)
ck[ck$conflict, ]
#>       id parent type markers opposite       rate conflict
#> 1 G1_001   F001 sire     300       23 0.07666667     TRUE
# one remedy: the parent in conflict becomes unknown
ruim <- ck[ck$conflict & ck$type == "sire", ]
ped$sire[match(ruim$id, ped$id)] <- "0"