impute_mv: Impute MV in data.

In tidymass/masscleaner: Data cleaning for LC-MS based metabolomics data

Impute MV in data.

Description

Impute MV in data.

Usage

impute_mv(
  object,
  sample_id,
  method = c("knn", "rf", "mean", "median", "zero", "minimum", "bpca", "svdImpute",
    "ppca"),
  k = 10,
  rowmax = 0.5,
  colmax = 0.8,
  maxp = 1500,
  rng.seed = 362436069,
  maxiter = 10,
  ntree = 100,
  decreasing = FALSE,
  nPcs = 2,
  maxSteps = 100,
  threshold = 1e-04,
  ...
)

Arguments

object	A mass_dataset object.
sample_id	which samples you want to impute missing value? It is a index or character vector (sample_id)
method	Imputation method. It contains "knn", "rf" (missForest), "mean", "median", "zero", "minium", "bpca" (BPCA), "svd" (SVD) and "ppca" (PPCA). Default is "knn". The detial of this method can be find in detail and reference paperes.
k	See ?impute.knn
rowmax	See ?impute.knn
colmax	See ?impute.knn
maxp	See ?impute.knn
rng.seed	See ?impute.knn
maxiter	See ?missForest
ntree	See ?missForest
decreasing	See ?missForest
nPcs	See ?bpca
maxSteps	See ?bpca
threshold	See ?bpca
...	Other arguments.

Value

A new mass_dataset object.

Author(s)

Xiaotao Shen [e-mail]

Examples

library(massdataset)
data("expression_data")
data("sample_info")
data("variable_info")
object =
  create_mass_dataset(
    expression_data = expression_data,
    sample_info = sample_info,
    variable_info = variable_info
  )
object

get_mv_number(object)
massdataset::get_mv_number(object, by = "sample")

###remove variables who have mv in more than 20% QC samples
qc_id =
  object %>%
  activate_mass_dataset(what = "sample_info") %>%
  filter(class == "QC") %>%
  pull(sample_id)

subject_id =
  object %>%
  activate_mass_dataset(what = "sample_info") %>%
  filter(class == "Subject") %>%
  pull(sample_id)

object =
  object %>%
  mutate_variable_na_freq(according_to_samples = qc_id) %>%
  mutate_variable_na_freq(according_to_samples = subject_id) %>%
  activate_mass_dataset(what = "variable_info") %>%
  filter(na_freq < 0.2 & na_freq.1 < 0.5)

###remove samples with MV > 50% except Blank samples
object =
  filter_samples(
    object = object,
    flist = function(x) {
      sum(is.na(x)) / nrow(object) < 0.5
    },
    apply_to = c(qc_id, subject_id),
    prune = TRUE
  )

blank_id =
  object %>%
  activate_mass_dataset(what = "sample_info") %>%
  filter(class == "Blank") %>%
  pull(sample_id)

object1 =
  impute_mv(object = object,
            sample_id = blank_id,
            method = "zero")

object1 %>%
  activate_mass_dataset(what = "expression_data") %>%
  select(dplyr::contains("Blank")) %>%
  extract_expression_data() %>%
  head()

object2 =
  impute_mv(object = object,
            sample_id = subject_id,
            method = "knn")

object2 %>%
  activate_mass_dataset(what = "sample_info") %>%
  filter(class == "Subject") %>%
  extract_expression_data() %>%
  head()

tidymass/masscleaner documentation built on Sept. 4, 2023, 3:21 a.m.