popkin_A: Compute popkin's 'A' and 'M' matrices from genotypes
In StoreyLab/popkin: Estimate Kinship and FST under Arbitrary Population Structure

popkin_A

R Documentation

Compute popkin's `A` and `M` matrices from genotypes

Description

This function returns lower-level, intermediate calculations for the main popkin function. These are not intended for most users, but rather for researchers studying the estimator.

Usage

popkin_A(
  X,
  n = NA,
  loci_on_cols = FALSE,
  mean_of_ratios = FALSE,
  mem_factor = 0.7,
  mem_lim = NA,
  m_chunk_max = 1000
)

Arguments

`X`	Genotype matrix, `BEDMatrix` object, or a function `X(m)` that returns the genotypes of all individuals at `m` successive locus blocks each time it is called, and `NULL` when no loci are left. If a regular matrix, `X` must have values only in `c(0, 1, 2, NA)`, encoded to count the number of reference alleles at the locus, or `NA` for missing data.
`n`	Number of individuals (required only when `X` is a function, ignored otherwise). If `n` is missing but `subpops` is not, `n` is taken to be the length of `subpops`.
`loci_on_cols`	If `TRUE`, `X` has loci on columns and individuals on rows; if `FALSE` (default), loci are on rows and individuals on columns. Has no effect if `X` is a function. If `X` is a `BEDMatrix` object, `loci_on_cols` is ignored (set automatically to `TRUE` internally).
`mean_of_ratios`	Chose how to weigh loci. If `FALSE` (default) loci have equal weights (in terms of variance, rare variants contribute less than common variants; also called the "ratio-of-means" version, this has known asymptotic behavior). If `TRUE`, rare variant loci are upweighed (in terms of variance, contributions are approximately the same across variant frequencies; also called the "mean-of-ratios" version, its asymptotic behavior is less well understood but performs better for association testing).
`mem_factor`	Proportion of available memory to use loading and processing data. Ignored if `mem_lim` is not `NA`.
`mem_lim`	Memory limit in GB, used to break up data into chunks for very large datasets. Note memory usage is somewhat underestimated and is not controlled strictly. Default in Linux is `mem_factor` times the free system memory, otherwise it is 1GB (Windows, OSX and other systems).
`m_chunk_max`	Sets the maximum number of loci to process at the time. Actual number of loci loaded may be lower if memory is limiting.

Value

A named list containing:

A: n-by-n matrix, for individuals j and k, of average w_i * ( ( x_ij - 1 ) * ( x_ik - 1 ) - 1) values across all loci i in X; if mean_of_ratios = FALSE, w_i = 1, otherwise w_i = 1 / (p_est_i * (1 - p_est_i) ) where p_est_i is the reference allele frequency.
M: n-by-n matrix of sample sizes (number of loci with non-missing individual j and k pairs, used to normalize A)

Examples

# Construct toy data
X <- matrix(c(0,1,2,1,0,1,1,0,2), nrow = 3, byrow = TRUE) # genotype matrix

# NOTE: for BED-formatted input, use BEDMatrix!
# "file" is path to BED file (excluding .bed extension)
# library(BEDMatrix)
# X <- BEDMatrix(file) # load genotype matrix object

obj <- popkin_A(X) # calculate A and M from genotypes
A <- obj$A
M <- obj$M

StoreyLab/popkin documentation built on Jan. 10, 2023, 1:39 p.m.