minor fix for namespace not loading within "do.call"

[Chunmingl]

No description provided.

[danielnachun]

If you can change:

args_list <- list(...)

to

args_list <- list2(...)

and then make this change

Suggested change

	fit.mr.ash <- mr.ash.alpha::mr.ash(X = X, y = y, sa2 = if (init_prior_sd) init_prior_sd(X, y)^2 else NULL, args_list)
	fit.mr.ash <- mr.ash.alpha::mr.ash(X = X, y = y, sa2 = if (init_prior_sd) init_prior_sd(X, y)^2 else NULL, !!!args_list)

It should work for the arguments

I think you'll also then need:

#' @importFrom !!! rlang
#' @importFrom list2 rlang

in the roxygen comment.

[Chunmingl]

I tried with the updated function below

mrash_weights <- function(X, y, init_prior_sd = TRUE, ...) {
  # Make sure mr.ash is installed
  if (! requireNamespace("mr.ash.alpha", quietly = TRUE)) {
    stop("To use this function, please install mr.ash: https://github.com/stephenslab/mr.ash.alpha")
  }
  args_list <- list2(...)
  if (!"beta.init" %in% names(args_list)) {
    args_list$beta.init <- lasso_weights(X, y)
  }
  fit.mr.ash <- mr.ash.alpha::mr.ash(X = X, y = y, 
              sa2 = if (init_prior_sd) init_prior_sd(X, y)^2 else NULL, !!!args_list)
  predict(fit.mr.ash, type = "coefficients")[-1]
}

and I got error message of

Error in `map()`:
i In index: 5.
Caused by error in `!args_list`:
! invalid argument type
Traceback:

initially I thought it was because the step that adds additional list of $beta.init but got the same error by unlist with !!! and list2 again after adding beta.init
args_list$beta.init <- lasso_weights(X, y)

[Chunmingl]

I tried as below and it seemed to work, please advise

mrash_weights <- function(X, y, init_prior_sd = TRUE, ...) {
  # Make sure mr.ash is installed
  if (! requireNamespace("mr.ash.alpha", quietly = TRUE)) {
    stop("To use this function, please install mr.ash: https://github.com/stephenslab/mr.ash.alpha")
  }
  args_list <- list2(...)
  if (!"beta.init" %in% names(args_list)) {
    args_list$beta.init <- lasso_weights(X, y)
  }
  fit.mr.ash <- do.call(getFromNamespace("mr.ash", "mr.ash.alpha"), 
                c(list(X = X, y = y, sa2 = if (init_prior_sd) init_prior_sd(X, y)^2 else NULL), args_list))
  predict(fit.mr.ash, type = "coefficients")[-1]
}

[danielnachun]

Okay that work around is actually fine, and you can switch list2 back to list.