Quantile-level RIF GWAS — quantile

Runs a genome-wide association scan at user-specified quantile levels (taus) using Recentered Influence Functions (RIF). Supports two covariate modes:

"residualize": partial out C from SNPs (but never from Y).
"include": include C directly in the regression alongside each SNP.

Usage

quantile_gwas(
  Y,
  G,
  taus = seq(0.1, 0.9, 0.05),
  C = NULL,
  covar_mode = c("residualize", "include"),
  density_floor = 1e-08,
  benchmark = TRUE,
  verbose = TRUE
)

Arguments

Y

Numeric vector (length N), phenotype.

G

Numeric matrix (N x P) of SNP dosages or genotypes.

taus

Numeric vector of quantile levels.

C

Optional N x K covariate matrix.

covar_mode

Either "residualize" (default) or "include".

"residualize": partial out C from G before regression.
"include": include C directly in regression.

density_floor

Positive scalar; lower bound for estimated densities.

benchmark

Logical; if TRUE, include timing info.

verbose

Logical; print progress messages.

Value

A list with taus, q_tau, fhat_tau, Q_slope, SE_tau, timing.