Expected likelihood ratio
This function computes the expected LR for a single marker, in a kinship test
comparing two hypothesised relationships between a set of individuals. The
true relationship may differ from both hypotheses. Some individuals may
already be genotyped, while others are available for typing. The
implementation uses oneMarkerDistribution() to find the joint genotype
distribution for the available individuals, conditional on the known data, in
each pedigree.
expectedLR(numeratorPed, denominatorPed, truePed = numeratorPed, ids, marker)
numeratorPed |
A |
denominatorPed |
A |
truePed |
A |
ids |
A vector of ID labels corresponding to untyped pedigree members. (These must be members of all three input pedigrees). |
marker |
either a marker object compatible with |
A positive number.
#---------
# Curious example showing that ELR may decrease
# by typing additional reference individuals
#---------
# Numerator ped
numPed = nuclearPed(father = "fa", mother = "mo", child = "ch")
# Denominator ped: fa, mo, ch are unrelated. Ugly hack!
denomPed = nuclearPed(father = "fa", mother = "mo", nch = 1)
denomPed = addChildren(denomPed, father = "ch", mother = "mo", nch = 1)
# Scenario 1: Only mother is typed; genotype 1/2
p = 0.9
m1 = marker(numPed, mo = 1:2, afreq = c("1" = p, "2" = 1-p))
expectedLR(numPed, denomPed, ids = "ch", marker = m1)
1/(8*p*(1-p)) + 1/2 # exact formula
# Scenario 2: Include father, with genotype 1/1
m2 = m1
genotype(m2, id = "fa") = c(1, 1)
expectedLR(numPed, denomPed, ids = "ch", marker = m2)
1/(8*p*(1-p)) + 1/(4*p^2) # exact formulaPlease choose more modern alternatives, such as Google Chrome or Mozilla Firefox.