compareCI: Obsolete. Compare kinship confidence intervals

In thoree/inbred: Different functions related to LR, bootstrap etc

Description

The purpose is to compare confidence intervals (CI-s) based on parametric bootstrap and nonparametric bootstrap. For a simple case, iid SNPs with frequency 0.5 and kappa2 = 0, the asymptotic counterparts are provided.

Usage

compareCI(
  theta,
  x,
  ids,
  n = NULL,
  N = 2,
  B = 2,
  seed = NULL,
  asymptotic = TRUE,
  bca = FALSE
)

Arguments

theta	Double kappa0.
x	ped object with allele frequencies.
ids	Id of pair.
n	Integer. No of markers, only used in the asymptoic case
N	Integer. No of simulations.
B	Integer. No of bootstraps.
seed	Integer.
bca	Logical.

Details

The equations for the simple asymptotic case are documented separately. CI-s for the bootstrap are calculated using the percentile method. Only estimates in the asymptotic case are constrained by kappa2 = 0. This makes comparison between the asymptotic and the bootstrap estimates a bit unfair.

Value

Returns a list with length 3 or 4 (if asymptotic is included). The last element of the list contains averaged CI, point estimate, coverage, and dist (as defined in 'ibdBootstrap'), the former elements contain the same information for each CI.

Examples

library(forrel)
library(pedprobr)

# Example 1. Assumptions for simple asymptotics met.
x = halfSibPed(1)
n = 100# no of markers
N = 100 # no of simulations
B = 1000 # no of  bootstraps
bca = FALSE
## Frequencies for bootstrap
freq = list()
for (i in 1:n)
 freq[[i]] =  list(alleles = 1:2, afreq = c(0.5, 0.5))
x = setMarkers(x, locusAttributes = freq)

## Parent offspring
ids = c(1, 4); theta = 0; seed = 17
foo1 = compareCI(theta, x, ids, n, N, B, bca = bca)

## half sibs
ids = c(4,5); theta = 0.5; seed = 17;
foo2 = compareCI(theta, x, ids, n, N, B, bca = bca, seed)

## Unrelated
ids = c(1,3); theta = 1
foo3 = compareCI(theta, x, ids, n, N, B, bca = bca)

cbind(theta = rep(c(0,0.5,1), each = 3),
      rbind(foo1$average, foo2$average, foo3$average))

# Example 2. Strong deviation from iid

n1 = 19 # no of id markers
n2 = 1
MAF = 0.001 #minor allele frequency
n = n1 + n2 # no of markers
N = 2 # no of simulations
B = 100 # no of  bootstraps

## Frequencies for bootstrap
freq = list()
for (i in 1:n1)
 freq[[i]] =  list(alleles = 1:2, afreq = c(0.5, 0.5))
for (i in (n1+1):n)
 freq[[i]] =  list(alleles = 1:2, afreq = c(MAF, 1-MAF))
x = setMarkers(x, locusAttributes = freq)


## Parent offspring
ids = c(1, 4); theta = 0; seed = 17
foo1 = compareCI(theta, x, ids, n, N, B, asymptotic = FALSE)

## half sibs
ids = c(4,5); theta = 0.5; seed = 17;
foo2 = compareCI(theta, x, ids, n, N, B, asymptotic= FALSE)

## Unrelated
ids = c(1,3); theta = 1
foo3 = compareCI(theta, x, ids, n, N, B, asymptotic = FALSE)

cbind(theta = rep(c(0,0.5,1), each = 2),
      rbind(foo1$average, foo2$average, foo3$average))

thoree/inbred documentation built on March 28, 2021, 7:42 p.m.