A function to choose a data-driven reference set.

chooseRefSet(
  fitted_model,
  reference_set_size,
  reference_set_covariate,
  constraint_fn
)

Arguments

fitted_model: The output from radEmu::emuFit when applied to data with a constraint over all taxa.
reference_set_size: The size of the reference set if it is data-driven, default is set to 50. We recommend a reference set of size 30-100 for the best balance of computational efficiency and estimation precision.
reference_set_covariate: If the reference set is data-driven, which covariates should it be chosen relative to. By default, this will be all covariates in the model (ignoring the intercept). However, if a model includes a main covariate that will be tested and several precision variables, we recommend choosing the reference set with respect to the main covariate of interest. This argument should be a vector of numbers that correspond to column indices in the X design matrix.
constraint_fn: The constraint function (by default the smoothed median).

Value

A list including the set of taxa of size reference_set_size with the smallest absolute L2 norm of estimated log fold-differences for the specified covariate(s), relative to the smoothed median log fold-difference over all taxa. Also a vector of length p (the number of columns in the design matrix) of differences between the constraint function over log fold-differences across all taxa and the constraint function log fold-difference over the reference set. Also the estimated B matrix, with each row shifted according to the constraint over the new reference set.