BayesN TWAS weights (Gaussian prior, ridge-equivalent)

Use Gaussian distribution as prior. Posterior means will be BLUP, equivalent to Ridge Regression.

Usage

bayesNWeights(
  X,
  y,
  Z = NULL,
  h2 = NULL,
  nit = 5000,
  nburn = 1000,
  nthin = 5,
  methodArgs = QggOptions()
)

Arguments

X

Numeric genotype / design matrix (samples x variants).

y

Numeric response (phenotype) vector of length nrow(X).

Z

Optional numeric matrix of fixed-effect covariates, or NULL.

h2

Numeric or NULL. Prior heritability; NULL leaves the engine's own default.

nit

Integer. Total MCMC iterations. Default 5000.

nburn

Integer. Burn-in iterations. Default 1000.

nthin

Integer. Thinning interval for retained MCMC samples. Default 5.

methodArgs

Options forwarded to qgg::gbayes, built with QggOptions. A bare list is refused: it cannot be checked against the engine, so a misspelled option would be silently ignored.

Value

A numeric vector of effect-size weights, one per variant (column of X); columns dropped for zero variance receive weight 0.

Examples

data(eqtlRegionExample)
X <- eqtlRegionExample$X[, 1:30]
y <- eqtlRegionExample$yRes
bayesNWeights(X, y)
#> Type of analysis performed: st-blr-individual-level-dense-ld
#>  [1] -0.035538371 -0.042740916  0.016500483 -0.005125461 -0.004186801
#>  [6] -0.010866823 -0.011351706  0.024922519  0.009398730  0.014700825
#> [11] -0.004026619 -0.042731837 -0.009060808  0.010570838 -0.008892798
#> [16]  0.029739587 -0.005919621 -0.008282481  0.009577588 -0.009052599
#> [21] -0.005856381 -0.012898004 -0.009089858 -0.002869922 -0.016079454
#> [26] -0.012692163 -0.035911203 -0.009195258  0.012264962  0.013961942