esmr: Empirical Shrinkage Multivariable MR
In jean997/mrScan: Empirical Shrinkage Mendelian Randomization

esmr	R Documentation

Empirical Shrinkage Multivariable MR

Description

Empirical Shrinkage Multivariable MR

esmr

Usage

esmr(
  beta_hat_X,
  se_X,
  beta_hat_Y = NULL,
  se_Y = NULL,
  G = NULL,
  R = NULL,
  pval_thresh = NULL,
  variant_ix = NULL,
  ld_scores = NULL,
  RE = NULL,
  tau_init = NULL,
  fix_tau = FALSE,
  ebnm_fn = flashier::flash_ebnm(prior_family = "point_normal", optmethod = "nlm"),
  g_init = NULL,
  fix_g = FALSE,
  max_iter = 100,
  sigma_beta = Inf,
  tol = "default",
  restrict_dag = TRUE,
  direct_effect_template = NULL,
  direct_effect_init = NULL,
  beta_joint = TRUE,
  augment_G = TRUE
)

Arguments

`beta_hat_X`	Matrix of SNP-exposure associations (p by K)
`se_X`	matrix of standard errors of beta_hat_X
`beta_hat_Y`	Vector of SNP-outcome associations (length p)
`se_Y`	Standard errors of beta_hat_Y
`G`	G matrix. If NULL, G will be estimated using the method given in g_type.
`R`	Optional correlation matrix for overlapping samples.
`pval_thresh`	p-value threshold for estimation
`variant_ix`	Instead of using pval_thresh, directly specify the indices of variants used for estimation.
`ebnm_fn`	Options prior distribution family. Defaults to point-normal.
`max_iter`	Maximum number of iterations
`sigma_beta`	Optional prior variance for causal parameters
`tol`	Convergence tolerance
`beta_joint`	Use joint updates for beta (suggest TRUE)
`augment_G`	Augment estimated G
`g_type`	Method to estimate G. Suggest "gfa"