simgl {hwep}

Simulator for genotype likelihoods.

Description

Uses the updog R package for simulating read counts and generating genotype log-likelihoods.

Usage

simgl(
  nvec,
  rdepth = 10,
  od = 0.01,
  bias = 1,
  seq = 0.01,
  ret = c("gl", "gp", "all"),
  est = FALSE,
  ...
)

Arguments

nvec

The genotype counts. nvec[k] contains the number of individuals with genotype k-1.

rdepth

The read depth. Lower means more uncertain.

od

The overdispersion parameter. Higher means more uncertain.

bias

The allele bias parameter. Further from 1 means more bias. Must greater than 0.

seq

The sequencing error rate. Higher means more uncertain.

ret

The return type. Should we just return the genotype likelihoods ("gl"), just the genotype posteriors ("gp"), or the entire updog output ("all")

est

A logical. Estimate the updog likelihood parameters while genotype (TRUE) or fix them at the true values (FALSE)? More realistic simulations would set this to TRUE, but it makes the method much slower.

...

Additional arguments to pass to flexdog_full().

Value

By default, a matrix. The genotype (natural) log likelihoods. The rows index the individuals and the columns index the dosage. So gl[i,j] is the genotype log-likelihood for individual i at dosage j - 1.

Author(s)

David Gerard

Examples

set.seed(1)
simgl(c(1, 2, 1, 0, 0), model = "norm", est = TRUE)
simgl(c(1, 2, 1, 0, 0), model = "norm", est = FALSE)