Skip to contents

Simulate Multiple Traits from Genotype and Effect Sizes

Source: R/simulate_linreg.R
sim_multi_traits.Rd

This function simulates multiple traits (phenotypes) based on genotype data, a matrix of effect sizes, and heritability. It allows specifying if the heritability is total or per-SNP, optionally scales the phenotypes, and can handle residual correlations. per eQTL heritability is 0.05 to 0.07 according to https://www.nature.com/articles/ng.3506

Usage

sim_multi_traits(
  G,
  B,
  h2g,
  is_h2g_total = TRUE,
  max_h2g = 1,
  residual_corr = NULL,
  null_sigma = sqrt(0.1)
)

Arguments

G

Genotype matrix.

B

Matrix of effect sizes for multiple traits.

h2g

Heritability (proportion of variance explained by genetics).

is_h2g_total

Logical indicating if h2g is total (TRUE) or per-SNP (FALSE).

max_h2g

Maximum heritability allowed (default 1).

residual_corr

Matrix of residual correlations (NULL for independent samples).

null_sigma

Standard deviation for null traits (default sqrt(0.1)).

Value

A list containing the simulated phenotypes matrix (t * n, t = trait number, n = sample size) (`P`) and residual variance (`residual_var`).

Examples

G = matrix(rbinom(1000, 2, 0.5), nrow = 1000, ncol = 50) 
# Simulating effect sizes for two traits
B = sim_beta(G, ncausal = 5, ntrait = 3, is_h2g_total = F, shared_pattern = "all")
P = sim_multi_traits(G, B, h2g = 0.1, is_h2g_total = T, max_h2g = 1)