simfam (version 1.1.6)

kinship_last_gen: Calculate kinship matrix for last generation of a pedigree with structured founders

Description

A wrapper around the more general kinship_fam(), specialized to save memory when only the last generation is desired (kinship_fam() returns kinship for the entire pedigree in a single matrix). This function assumes that generations are non-overlapping (met by the output of sim_pedigree()), in which case each generation g can be drawn from generation g-1 data only. That way, only two consecutive generations need be in memory at any given time. The partitioning of individuals into generations is given by the ids parameter (again matches the output of sim_pedigree()).

Usage

kinship_last_gen(kinship, fam, ids, missing_vals = c("", 0))

Value

The kinship matrix of the last generation (the intersection of ids[ length(ids) ] and fam$id). The columns/rows of this matrix are last-generation individuals in the order that they appear in fam$id.

Arguments

kinship

The kinship matrix of the founders. This matrix must have column and row names that identify each founder (matching codes in fam$id). Individuals may be in a different order than fam$id. Extra individuals in kinship but absent in fam$id will be silently ignored. A traditional pedigree calculation would use kinship = diag(n)/2 (plus appropriate column/row names), where n is the number of founders, to model unrelated and outbred founders. However, if kinship measures the population kinship estimates between founders, the output is also a population kinship matrix (which combines the structural/ancestral and local/pedigree relatedness values into one).

fam

The pedigree data.frame, in plink FAM format. Only columns id, pat, and mat are required. id must be unique and non-missing. Founders must be present, and their pat and mat values must be missing (see below). Non-founders must have both their parents be non-missing. Parents must appear earlier than their children in the table.

ids

A list containing vectors of IDs for each generation. All these IDs must be present in fam$id. If IDs in fam and ids do not fully agree, the code processes the IDs in the intersection, which is helpful when fam is pruned but ids is the original (larger) set.

missing_vals

The list of ID values treated as missing. NA is always treated as missing. By default, the empty string ('') and zero (0) are also treated as missing (remove values from here if this is a problem).

See Also

Plink FAM format reference: https://www.cog-genomics.org/plink/1.9/formats#fam

Examples

# A small pedigree, two parents and two children.
# A minimal fam table with the three required columns.
# Note "mother" and "father" have missing parent IDs, while children do not
library(tibble)
fam <- tibble(
  id = c('father', 'mother', 'child', 'sib'),
  pat = c(NA, NA, 'father', 'father'),
  mat = c(NA, NA, 'mother', 'mother')
)
# need an `ids` list separating the generations
ids <- list( c('father', 'mother'), c('child', 'sib') )

# Kinship of the parents, here two unrelated/outbred individuals:
kinship <- diag(2)/2
# Name the parents with same codes as in `fam`
# (order can be different)
colnames( kinship ) <- c('mother', 'father')
rownames( kinship ) <- c('mother', 'father')
# For a clearer example, make the father slightly inbred
# (a self-kinship value that exceeds 1/2):
kinship[2,2] <- 0.6

# calculate the kinship matrix of the children
kinship2 <- kinship_last_gen( kinship, fam, ids )
kinship2