Evaluates imputations methods by NRMSE or AUC

benchmark_imputation(
  scm = NULL,
  assay = "score",
  sparse_prop = seq(0.1, 0.9, 0.1),
  iterations = 3,
  imp_methods = c(iPCA = function(...) impute_regions(type = "iPCA", ...), RF =
    function(...) impute_regions(type = "RF", ...), kNN = function(...)
    impute_regions(type = "kNN", ...)),
  type = "RMSE"
)

Arguments

scm	`scMethrix`; the single cell methylation experiment
assay	string; name of an existing assay. Default = "score"
sparse_prop	numeric; A sparsity proportion between 0 and 1. E.g. 0.1 replaces 10% of the matrix with NA
iterations	integer; Number of iterations to test
imp_methods	closure; The imputation methods to compare.
type	character; descriptive statistic. Can be either "AUC" or "RMSE". Default "RMSE"

Value

ggplot; The graph showing the NRMSE for each imputation method at each sparsity

Details

Does stuff

Examples

data('scMethrix_data')
if (FALSE) {
scMethrix_data <- impute_regions(scMethrix_data, new_assay="impute",type="RF")
benchmark_imputation(scMethrix_data, assay="impute", sparse_prop = c(0.1,0.5,0.85))
}