simulate_mr_mash_data: Simulate data to test 'mr.mash'.

In stephenslab/mr.mash.alpha: Multiple Regression with Multivariate Adaptive Shrinkage

Simulate data to test mr.mash.

Description

Function to simulate data from MN_{nxr}(XB, I, V), where X \sim N_p(0, Gamma), B \sim \sum_k w_k N_r(0, Sigma_k), with Gamma, w_k, Sigma_k, and V defined by the user.

Usage

simulate_mr_mash_data(
  n,
  p,
  p_causal,
  r,
  r_causal = list(1:r),
  intercepts = rep(1, r),
  pve = 0.2,
  B_cor = 1,
  B_scale = 1,
  w = 1,
  X_cor = 0,
  X_scale = 1,
  V_cor = 0
)

Arguments

n	scalar indicating the number of samples.
p	scalar indicating the number of variables.
p_causal	scalar indicating the number of causal variables.
r	scalar indicating the number of responses.
r_causal	a list of numeric vectors (one element for each mixture component) indicating in which responses the causal variables have an effect.
intercepts	numeric vector of intercept for each response.
pve	per-response proportion of variance explained by the causal variables.
B_cor	scalar or numeric vector (one element for each mixture component) with positive correlation [0, 1] between causal effects.
B_scale	scalar or numeric vector (one element for each mixture component) with the diagonal value for Sigma_k;
w	scalar or numeric vector (one element for each mixture component) with mixture proportions associated to each mixture component.
X_cor	scalar indicating the positive correlation [0, 1] between variables.
X_scale	scalar indicating the diagonal value for Gamma.
V_cor	scalar indicating the positive correlation [0, 1] between residuals

Value

A list with some or all of the following elements:

X	n x p matrix of variables.
Y	n x r matrix of responses.
B	p x r matrix of effects.
V	r x r residual covariance matrix among responses.
Sigma	list of r x r covariance matrices among the effects.
Gamma	p x p covariance matrix among the variables.
intercepts	r-vector of intercept for each response.
causal_responses	a list of numeric vectors of indexes indicating which responses have causal effects for each mixture component.
causal_variables	p_causal-vector of indexes indicating which variables are causal.
causal_vars_to_mixture_comps	p_causal-vector of indexes indicating from which mixture components each causal effect comes.

Examples

set.seed(1)
dat <- simulate_mr_mash_data(n=50, p=40, p_causal=20, r=5,
                             r_causal=list(1:2, 3:4), intercepts=rep(1, 5),
                             pve=0.2, B_cor=c(0, 1), B_scale=c(0.5, 1),
                             w=c(0.5, 0.5), X_cor=0.5, X_scale=1, 
                             V_cor=0)
                             
                             

stephenslab/mr.mash.alpha documentation built on Feb. 7, 2025, 10:06 p.m.