DNAprofiles (version 0.3.1)

ki.cdf: Computes CDF of KI between case profile and profile with stated relationship

Description

Computes the Cumulative Distribution Function of a Kinship Index (KI) comparing hypotheses hyp.1 vs hyp.2 for profiles with a given relationship (hyp.true) to the case profile (e.g. "FS" for full siblings).

Usage

ki.cdf(x, hyp.1, hyp.2 = "UN", hyp.true = "UN", freqs.ki = get.freqs(x), freqs.true = freqs.ki, markers = {     if (missing(x))          intersect(names(freqs.ki), names(freqs.true))     else get.markers(x) }, theta.ki = 0, theta.true = theta.ki, n.max = 1e+06)

Arguments

x

(optional) An integer matrix specifying a single profile.

hyp.1

A character string giving the hypothesis in the numerator of the $KI$. Should be one of ibdprobs, e.g. "FS" (full sibling) or "PO" (parent/offspring) or "UN" (unrelated).

hyp.2

A character string giving the hypothesis in the denominator of the $KI$. Should be one of ibdprobs, e.g. "FS" (full sibling) or "PO" (parent/offspring) or "UN" (unrelated). Defaults to "UN".

hyp.true

A character string specifying the true relationship between the case profile and the other profile. Should be one of ibdprobs, e.g. "FS" (full sibling) or "PO" (parent/offspring) or "UN" (unrelated). Defaults to "UN".

freqs.ki

A list specifying the allelic frequencies that are used when computing the $KI$.

freqs.true

(optionally) A list specifying the allelic frequencies that are used for computing the probabily distribution of the $KI$ under hyp.true. When not provided, the function will use freqs. One might use different allelic frequencies freqs.rel when for example the case profile and relative come from some population, while $KI$s are computed with frequencies from another population.

markers

Character vector stating the markers for which the KI distribution is derived. By default equal to all markers of x, or, if x is missing, to the intersection of the markers of freqs.ki and freqs.true.

theta.ki

numeric value specifying the amount of background relatedness.

theta.true

numeric value specifying the amount of background relatedness.

n.max

Maximum number of events stored in memory. See dists.product.pair for details.

Examples

# for one profile, obtain the CDF of the SI,
# both for true sibs and unrelated profiles
data(freqsNLsgmplus)

# sample a profile
x <- sample.profiles(N=1,freqsNLsgmplus)

cdf.fs <- ki.cdf(x,hyp.1="FS",hyp.true="FS")
cdf.un <- ki.cdf(x,hyp.1="FS",hyp.true="UN")

# the cdf's are *functions*
cdf.fs(1)
cdf.un(1)

# we also obtain an ROC curve easily
t <- 10^(seq(from=-10,to=10,length=100)) # some thresholds
fpr <- cdf.un(t,exc.prob=TRUE)
tpr <- cdf.fs(t,exc.prob=TRUE)

plot(log10(fpr),tpr,type="l")