Skip to contents

Compute mr.mash-RSS TWAS weights from summary statistics

Source: R/regularized_regression.R
mrmash_rss_weights.Rd

Multi-context summary-statistics analog of mrmash_weights: extracts coefficients from an existing mr.mashr::mr.mash.rss fit, or fits one from stat (variants x conditions) and LD.

Usage

mrmash_rss_weights(
  stat,
  LD,
  mrmash_rss_fit = NULL,
  data_driven_prior_matrices = NULL,
  canonical_prior_matrices = TRUE,
  S0 = NULL,
  w0 = NULL,
  V = NULL,
  covY = NULL,
  retain_fit = FALSE,
  ...
)

Arguments

stat

A list with z (variants x conditions matrix) and n (per-context numeric vector or scalar). May also include Bhat, Shat matrices.

LD

LD correlation matrix.

mrmash_rss_fit

Optional pre-fitted mr.mash.rss object; skips fitting when supplied.

data_driven_prior_matrices

Optional list with element U (list of raw covariance matrices). Passed directly to mr.mashr::expand_covs() alongside the canonical mixture.

canonical_prior_matrices

Logical. When TRUE (default), include the standard canonical mixture from mr.mashr::compute_canonical_covs(). When FALSE, data_driven_prior_matrices must be supplied.

S0

Optional pre-built list of prior covariance matrices, bypassing the canonical / data-driven construction.

w0

Optional prior mixture weights; defaults to compute_w0(Bhat, length(S0)).

V

Optional residual covariance matrix (K x K). When NULL, defaults to the identity matrix of size K.

covY

Optional response covariance matrix (K x K). When NULL, defaults to the identity matrix of size K.

retain_fit

If TRUE, attaches the fitted object as the "fit" attribute on the returned weights.

...

Additional arguments forwarded to mr.mashr::mr.mash.rss.

Value

A numeric matrix of per-variant per-context weights (variants x conditions).

Details

Follows the *_rss_weights(stat, LD, ...) contract. Expects stat$z to be a numeric matrix (variants x conditions) and stat$n a per-context numeric vector or scalar. stat$Bhat and stat$Shat are used if present; otherwise derived from Z and n.

Prior construction reuses the same infrastructure as the individual-level mrmash_wrapper: compute_grid + mr.mashr::compute_canonical_covs() + mr.mashr::expand_covs() for S0, and compute_w0 for the mixture weights. Supply data_driven_prior_matrices (e.g. from compute_cov_flash / compute_cov_diag) to add data-driven covariance components alongside the canonical mixture.