Skip to contents

Impute proteomics data

Source: R/Imputation.R
impute_proteomics.Rd

Complete imputation pipeline with 20 methods. Four pathways:

Usage

impute_proteomics(
  se,
  normalized_assay_name = "cycloess",
  imputed_assay_name = NULL,
  imp_method = "combo",
  mar_method = "Impseqrob",
  mnar_method = "min",
  prop_na_mnar = 0.51,
  prop_present_mar = 0.5,
  min_present_mar = 1,
  require_n_conditions = 1,
  max_na_prop = NULL,
  method_args = list(),
  with_value = NA_real_,
  verbose = TRUE
)

Arguments

se

SummarizedExperiment with normalized assay

normalized_assay_name

Name of the normalized assay to use (default: "cycloess")

imputed_assay_name

Name for the imputed assay. If NULL, it is built from the method: combo gives <mar_method>_<mnar_method>, softHybrid gives softHybrid_<mar_method>_<mnar_method>, and any single method gives its own name.

imp_method

Imputation method (default: "combo"). One of: "combo", "softHybrid", "bpca", "knn", "mice", "missForest", "Impseq", "Impseqrob", "QRILC", "MLE", "MinDet", "MinProb", "PI", "min", "halfmin", "zero", "nbavg", "with", "limpa", "none". "halfmin" is the DIA-NN half-minimum recipe: since the assay is on the log2 scale, halving the intensity means one unit below the global observed minimum.

mar_method

MAR method for combo/softHybrid mode (default: "Impseqrob")

mnar_method

MNAR method for combo/softHybrid mode (default: "min")

prop_na_mnar

NA proportion threshold for MNAR classification (default: 0.51)

prop_present_mar

Present proportion for MAR (default: 0.5)

min_present_mar

Minimum present values for MAR (default: 1)

require_n_conditions

Number of conditions required with presence (default: 1)

max_na_prop

Maximum NA proportion above which a protein is removed before imputation. NULL (the default) disables the filter, so every method receives the same proteins.

method_args

Named list of per-method argument lists (e.g., list(knn = list(k = 10), bpca = list(nPcs = 3)))

with_value

Constant value for imp_method="with" (default: NA_real_)

verbose

Print progress messages (default: TRUE)

Value

List with:

  • se: SummarizedExperiment with imputed assay added

  • x_imputed: Raw imputed matrix (for direct export)

  • prefilter_summary: Pre-filtering summary

  • imputation_summary: Imputation statistics

  • mnar_mask: MNAR mask matrix (combo only, NULL otherwise)

  • mar_mask: MAR mask matrix (combo only, NULL otherwise)

Details

  • imp_method = "none": no imputation

  • imp_method = "combo": two-stage MAR+MNAR with binary classification

  • imp_method = "softHybrid": sigmoid-weighted MAR+MNAR blend (continuous)

  • Any other method: apply to all NAs (no MAR/MNAR distinction)

Examples

data(nadia_dia)
norm <- normalize_proteomics(nadia_dia, norm_method = "cycloess",
                             verbose = FALSE)

# Default: two-stage combo, Impseqrob for MAR and min for MNAR
imp <- impute_proteomics(norm$se, normalized_assay_name = "cycloess",
                         verbose = FALSE)
SummarizedExperiment::assayNames(imp$se)
#> [1] "raw"           "log2"          "cycloess"      "Impseqrob_min"
sum(is.na(SummarizedExperiment::assay(imp$se, "Impseqrob_min")))
#> [1] 0

# Any single method works too, and per-method options go in method_args
imp2 <- impute_proteomics(norm$se, normalized_assay_name = "cycloess",
                          imp_method = "knn",
                          method_args = list(knn = list(k = 15)),
                          verbose = FALSE)
SummarizedExperiment::assayNames(imp2$se)
#> [1] "raw"      "log2"     "cycloess" "knn"