simulate_plots: Simulate a multi-environment breeding trial
In etnite/bwardr: Brian Ward personal R package

simulate_plots

R Documentation

Simulate a multi-environment breeding trial

Description

Simulate a multi-environment breeding trial

Usage

simulate_plots(
  X,
  snps = 1000,
  inds = 100,
  qtls = 20,
  n_envs = 4,
  n_reps = 2,
  var_phen = 10,
  h2 = 0.75,
  var_env = 5,
  var_rep = 0.5,
  beta_ab = 5,
  return_scaled = FALSE
)

Arguments

`X`	A Marker matrix in minor-allele dosage format, with individuals in rows and SNPs in columns
`snps`	Either an integer n, in which case n SNPs are randomly sampled out of X, or a vector of SNP indices, or a vector of SNP names. Setting to "all" will use all SNPs
`inds`	Either an integer n, in which case n individuals are randomly sampled out of X, or a vector of individual indices, or a vector of individual names. Setting to "all" will use all individuals
`qtls`	Either an integer n, in which case n evenly-spaced SNPs will be chosen as QTLs, or a vector of SNP indices to assign as QTLs, or vector of SNP names to assign as QTLs
`n_envs`	Integer - number of environments to simulate
`n_reps`	Integer - number of replicates within each environment to simulate
`var_phen`	Positive float - Phenotypic variance within each environment
`h2`	Proper fraction - Within-environment narrow-sense heritability value
`var_env`	Positive float - Variance between environments
`var_rep`	Positive float - Variance between replications within same environment
`beta_ab`	Improper fraction - Beta distribution shape parameter to control GxE effect. See details.
`return_scaled`	Logical - Indicates whether to scale the genetic signal and the error by the phenotypic standard deviation before returning the outputted plots data

Details

QTL effects are allowed to vary across environments by sampling out of a symmetric beta distribution. This implies that genotype-by-environment (GxE) interaction decreases as the α and β shape parameters of the distribution increase. At the limits, setting α = β = 1 makes the beta distribution equivalent to a uniform distribution - QTL effects may vary without any central tendency. Alternatively, setting α = β = Inf will set QTL effects constant across environments. Setting n_reps to 1 will simulate within-environment means. In this case, the var_rep value will have no effect.