multiPersonIBD: Multi-person IBD coefficients

In ribd: Pedigree-based Relatedness Coefficients

Multi-person IBD coefficients

Description

Computes the probabilities (coefficients) of all possible patterns of identity by descent (IBD) sharing at a single locus, among N>1 non-inbred members of a pedigree. The reported coefficients are "condensed" in the sense that allele ordering within each individual is ignored. For N = 2, the result should agree with the traditional "kappa" coefficients, as computed by kappaIBD(). This function is under development, and should be regarded as experimental. For now, the only cases handled are those with: N = 2 or 3, autosomal locus.

Usage

multiPersonIBD(x, ids, complete = FALSE, verbose = FALSE)

Arguments

x	A ped object.
ids	A vector of ID labels.
complete	A logical. If FALSE, only IBD patterns with nonzero probability are included in the output.
verbose	A logical. If TRUE, some computational details are printed.

Details

Consider N members of a pedigree, i1, i2, ... iN. A pattern of IBD sharing between these individuals is a sequence of N ordered pairs of labels, (a1_1, a1_2), (a2_1, a2_2), ... (aN_1, aN_2), where ai_1 and ai_2 represent the paternal and maternal allele of individual i, respectively. Equality of labels means that the corresponding alleles are IBD, and vice versa.

We say that two IBD patterns are equivalent if one can be transformed into the other by some combination of

• renaming the labels (without changing the structure)

• swapping the paternal/maternal labels of some individuals

Each equivalence class has a "minimal" element, using integer labels, and being minimal with respect to standard sorting. For example, the minimal element equivalent to ⁠(a,c),(d,c),(b,b)⁠ is ⁠(1,2),(2,3),(4,4)⁠.

Value

A data frame in which each row corresponds to an equivalence class of multi-person IBD patterns. The first column gives the calculated probability, followed by one column for each ids individual, describing the minimal element of the equivalence class. (See Details.) If complete = FALSE (the default) rows with probability 0 are removed.

Examples

### Trivial example: Trio ###
x = nuclearPed(1)
ids = 1:3
multiPersonIBD(x, ids, complete = TRUE)

### Example due to Peter Green ###
# Three (pairwise) cousins arranged in two different ways,
# with different 3-way IBD coefficients.

threeCousins1 = ped(
  id  = c('gf','gm','gf1','gf2','gf3','gm1','gm2','gm3',
          'f1','f2','f3','m1','m2','m3','c1','c2','c3'),
  fid = c(0,0,0,0,0,0,0,0,'gf1','gf2','gf3','gf','gf','gf',
          'f1','f2','f3'),
  mid = c(0,0,0,0,0,0,0,0,'gm1','gm2','gm3','gm','gm','gm',
          'm1','m2','m3'),
  sex = c(1,2,1,1,1,2,2,2,1,1,1,2,2,2,1,1,1))

threeCousins2 = ped(
  id  = c('gf1','gf2','gf3','gm1','gm2','gm3','f1','f2','f3',
          'm1','m2','m3','c1','c2','c3'),
  fid = c(0,0,0,0,0,0,'gf2','gf3','gf1','gf3','gf1','gf2',
          'f1','f2','f3'),
  mid = c(0,0,0,0,0,0,'gm2','gm3','gm1','gm3','gm1','gm2',
          'm1','m2','m3'),
  sex = c(1,1,1,2,2,2,1,1,1,2,2,2,1,1,1))

ids = c('c1','c2','c3')
multiPersonIBD(threeCousins1, ids)
multiPersonIBD(threeCousins2, ids)

ribd documentation built on May 29, 2024, 8:43 a.m.