forrel (version 1.6.1)

missingPersonLR: Likelihood ratio calculation for missing person identification

Description

This is a wrapper function for kinshipLR() for the special case of missing person identification. A person of interest (POI) is matched against a reference dataset containing genotypes of relatives of the missing person.

Usage

missingPersonLR(reference, missing, poi = NULL, verbose = TRUE, ...)

Value

The LRresult object returned by kinshipLR(), but without the trivial H2:H2 comparison.

Arguments

reference

A ped object with attached markers.

missing

The ID label of the missing member of reference.

poi

A singleton object, or NULL. If NULL, and missing is genotyped, this data is extracted and used as poi.

verbose

A logical.

...

Optional parameters to be passed on to kinshipLR().

Details

Note that this function accepts two forms of input:

1. With poi a typed singleton. This is the typical use case, when you want to compute the LR for some person of interest.

2. With poi = NULL, but missing being genotyped. The data for missing is then extracted as a singleton POI. This is especially useful in simulation procedures, e.g., for simulating the LR distribution of the true missing person.

See Examples for illustrations of both cases.

Examples

#------------------------------------------------
# Example: Identification of a missing grandchild
#------------------------------------------------

# Database with 5 STR markers (increase to make more realistic)
db = NorwegianFrequencies[1:5]

# Pedigree with missing person (MP); grandmother is genotyped
x = linearPed(2) |>
  relabel(old = 5, new = "MP") |>
  profileSim(markers = db, ids = "2", seed = 123)


### Scenario 1: Unrelated POI --------------------

# Generate random unrelated profile
poi = singleton("POI") |>
  profileSim(markers = db, seed = 1234)

# Compute LR
lr = missingPersonLR(x, missing = "MP", poi = poi)
lr
lr$LRperMarker


### Scenario 2: POI is the missing person --------
# A small simulation example

# Simulate profiles for MP conditional on the grandmother
N = 10
y = profileSim(x, N = N, ids = "MP", seed = 12345)

# Compute LRs for each sim
LRsims = lapply(y, missingPersonLR, missing = "MP", verbose = FALSE)

# Plot distribution
LRtotal = sapply(LRsims, function(a) a$LRtotal)
plot(density(LRtotal))

# LRs for each marker
LRperMarker = sapply(LRsims, function(a) a$LRperMarker)
LRperMarker

# Overlaying marker-wise density plots (requires tidyverse)
# library(tidyverse)
# t(LRperMarker) |> as_tibble() |> pivot_longer(everything()) |>
#   ggplot() + geom_density(aes(value, fill = name), alpha = 0.6)