kinshipLR {forrel}

Likelihood ratios for kinship testing

Description

This function computes likelihood ratios (LRs) for a list of pedigrees. One of the pedigrees (the last one, by default) is designated as 'reference', to be used in the denominator in all LR calculations. To ensure that all pedigrees use the same data set, one of the pedigrees may be chosen as 'source', from which data is transferred to all the other pedigrees.

Usage

kinshipLR(
  ...,
  ref = NULL,
  source = NULL,
  markers = NULL,
  linkageMap = NULL,
  keepMerlin = NULL,
  verbose = FALSE
)

Arguments

...

Pedigree alternatives. Each argument should be either a single ped object or a list of such. The pedigrees may be named; otherwise they are assigned names "H1", "H2", ... automatically.

It is also possible to pass a single list containing all the pedigrees.

ref

An index or name indicating which of the input pedigrees should be used as "reference pedigree", i.e., used in the denominator of each LR. If NULL (the default), the last pedigree is used as reference.

source

An index or name designating one of the input pedigrees as source for marker data. If given, marker data is transferred from this to all the other pedigrees (replacing any existing markers). The default action (source = NULL) is as follows: If all pedigree have attached markers, no transfers are done. If exactly one of the pedigrees have attached markers, these are transferred to the others. all other cases give an error.

markers

A vector of marker names or indices indicating which markers should be included. If NULL (the default) all markers are used.

linkageMap

If this is non-NULL, the markers are interpreted as being linked, and likelihoods will be computed by an external call to MERLIN.

The supplied object should be either:

• a data frame, whose first three columns must be (i) chromosome (ii) marker name (iii) centiMorgan position, or

• a map object created with ibdsim2::uniformMap() or ibdsim2::loadMap(). This will internally be applied to the attached markers to produce a suitable data frame as above.

keepMerlin

Either NULL (default) or the path to an existing folder. If given, MERLIN files are stored here, typically for debugging purposes.

verbose

A logical.

Details

By default, all markers are assumed to be unlinked. To accommodate linkage, a genetic map may be supplied with the argument linkageMap. This requires the software MERLIN to be installed.

Value

A LRresult object, which is essentially a list with entries

• LRtotal : A vector of length L, where L is the number of input pedigrees. The i'th entry is the total LR (i.e., the product over all markers) comparing pedigree i to the reference pedigree. The entry corresponding to the reference will always be 1.

• LRperMarker : A numerical matrix, where the i'th column contains the marker-wise LR values comparing pedigree i to the reference. The product of all entries in a column should equal the corresponding entry in LRtotal.

• likelihoodsPerMarker : A numerical matrix of the same dimensions as LRperMarker, but where the entries are likelihood of each pedigree for each marker.

• time : Elapsed time

Author(s)

Magnus Dehli Vigeland and Thore Egeland

See Also

LRpower(), pedprobr::likelihoodMerlin()

Examples

### Example 1: Full vs half sibs

# Simulate 5 markers for a pair of full sibs
ids = c("A", "B")
sibs = nuclearPed(children = ids)
sibs = simpleSim(sibs, N = 5, alleles = 1:4, ids = ids, seed = 123)

# Create two alternative hypotheses
halfsibs = relabel(halfSibPed(), old = 4:5, new = ids)
unrel = list(singleton("A"), singleton("B"))

# Compute LRs. By default, the last ped is used as reference
kinshipLR(sibs, halfsibs, unrel)

# Input pedigrees can be named, reflected in the output
kinshipLR(S = sibs, H = halfsibs, U = unrel)

# Select non-default reference (by index or name)
kinshipLR(S = sibs, H = halfsibs, U = unrel, ref = "H")

# Alternative syntax: List input
peds = list(S = sibs, H = halfsibs, U = unrel)
kinshipLR(peds, ref = "H", source = "S", verbose = TRUE)

# Detailed results
res = kinshipLR(peds)
res$LRperMarker
res$likelihoodsPerMarker


### Example 2: Separating grandparent/halfsib/uncle-nephew

# Requires ibdsim2 and MERLIN
if(requireNamespace("ibdsim2", quietly = TRUE) && pedprobr::checkMerlin()) {

  # Load recombination map
  map = ibdsim2::loadMap("decode19", uniform = TRUE)   # unif for speed

  # Define pedigrees
  ids = c("A", "B")
  H = relabel(halfSibPed(),   old = c(4,5), new = ids)
  U = relabel(avuncularPed(), old = c(3,6), new = ids)
  G = relabel(linearPed(2),   old = c(1,5), new = ids)

  # Attach FORCE panel of SNPs to G
  G = setSNPs(G, FORCE[1:10, ])  # use all for better results

  # Simulate recombination pattern in G
  ibd = ibdsim2::ibdsim(G, N = 1, ids = ids, map = map)

  # Simulate genotypes conditional on pattern
  G = ibdsim2::profileSimIBD(G, ibdpattern = ibd)

  # Compute LR (genotypes are automatically transferred to H and U)
  kinshipLR(H, U, G, linkageMap = map)
}