Parallel Subset-Based Interaction Effect Analysis — subset_perm_interaction

Runs stratified random subsets through a permutation-based interaction effect pipeline in parallel using C++ core logic. Uses future and furrr for parallelization with automatic plan handling.

Usage

subset_perm_interaction_effect(
  X,
  Y,
  MX,
  MY,
  g_col,
  n_iter = 1000,
  B = 1000,
  save_null = FALSE,
  seed = NULL,
  workers = future::availableCores() - 1
)

Arguments

X: Data frame or matrix of predictors.
Y: Outcome variable or matrix.
MX: Covariates for the test split.
MY: Covariates for the inference split.
g_col: Name of the grouping or genotype column.
n_iter: Number of random subsets to run. Default: 1000.
B: Number of permutations per subset. Default: 1000.
save_null: Store full permutation null matrix? Default: FALSE.
seed: Optional seed for reproducibility. Default: NULL.
workers: Number of parallel workers. Default: availableCores()-1.

Value

A list with aggregated stats, iteration stats, sample log, permutation nulls (optional), and meta information.

Details

The function sets a parallel plan with `multisession` and restores the previous plan automatically.