1. qmtools
  2. Commit 492d166bb9e4d7d7ab03c8838b3aad51a9c7c86b
Browse code

Added manual for poplin_impute.

Jaehyun Joo authored on 29/11/2021 05:59:07
Showing 12 changed files
NAMESPACE
History View file @ 492d166
... ... 
@@ -64,6 +64,7 @@ export(poplin_impute_mean)
64 64 
 export(poplin_impute_median)
65 65 
 export(poplin_impute_pca)
66 66 
 export(poplin_impute_randomforest)
67 
+export(poplin_impute_simple)
67 68 
 export(poplin_naplot)
68 69 
 export(poplin_normalize)
69 70 
 export(poplin_normalize_auto)
R/AllGenerics.R
History View file @ 492d166
... ... 
@@ -231,6 +231,12 @@ setGeneric(
231 231 
   function(x, ...) standardGeneric("poplin_impute_mean")
232 232 
 )
233 233
234 
+##' @export
235 
+setGeneric(
236 
+  "poplin_impute_simple",
237 
+  function(x, ...) standardGeneric("poplin_impute_simple")
238 
+)
239 
+
234 240 
 ##' @export
235 241 
 setGeneric(
236 242 
   "poplin_impute_pca",
R/imputation-functions.R
History View file @ 492d166
... ... 
@@ -1,19 +1,64 @@
1 1 
 .poplin_impute <- function(x,
2 
-                           method = c("knn", "halfmin", "median",
3 
-                                      "mean", "pca", "randomforest"),
2 
+                           method = c("knn", "randomforest", "pca", "simple"),
4 3 
                            ...) {
5 4 
   method <- match.arg(method)
6 5 
   switch(
7 6 
     method,
8 7 
     knn = .poplin_impute_knn(x, ...),
9 
-    halfmin = .poplin_impute_halfmin(x, ...),
10 
-    median = .poplin_impute_median(x, ...),
11 
-    mean = .poplin_impute_mean(x, ...),
12 
-    pca = .poplin_impute_pca(x, ...),
13 8 
     randomforest = .poplin_impute_randomforest(x, ...),
9 
+    pca = .poplin_impute_pca(x, ...),
10 
+    ## halfmin = .poplin_impute_halfmin(x, ...),
11 
+    ## median = .poplin_impute_median(x, ...),
12 
+    ## mean = .poplin_impute_mean(x, ...),
13 
+    simple = .poplin_impute_simple(x, ...)
14 14 
     )
15 15 
 }
16 16
17 
+## Knn imputation
18 
+.poplin_impute_knn <- function(x, by = c("feature", "sample"), ...) {
19 
+  if (!requireNamespace("VIM", quietly = TRUE)) {
20 
+    stop("Package 'VIM' is required. Please install and try again.")
21 
+  }
22 
+  by <- match.arg(by)
23 
+  if (by == "feature") {
24 
+    out <- VIM::kNN(x, ...)[, 1:ncol(x)]
25 
+    ## VIM package internally converts x as data.table, which drops rownames
26 
+    rownames(out) <- rownames(x)
27 
+  } else {
28 
+    out <- t(VIM::kNN(t(x), ...))[1:nrow(x), ]
29 
+    colnames(out) <- colnames(x)
30 
+  }
31 
+  as.matrix(out)
32 
+}
33 
+
34 
+## Random forest imputation
35 
+.poplin_impute_randomforest <- function(x, ...) {
36 
+  if (!requireNamespace("missForest", quietly = TRUE)) {
37 
+    stop("Package 'missForest' is required. Please install and try again.")
38 
+  }
39 
+  t(missForest::missForest(t(x), ...)$ximp)
40 
+}
41 
+
42 
+## Bayesian PCA imputation
43 
+.poplin_impute_pca <- function(x, type = c("bpca", "ppca", "nipals", "svdImpute"), ...) {
44 
+  if (!requireNamespace("pcaMethods", quietly = TRUE)) {
45 
+    stop("Package 'pcaMethods' is required. Please install and try again.")
46 
+  }
47 
+  type <- match.arg(type)
48 
+  t(pcaMethods::pca(t(x), method = type, ...)@completeObs)
49 
+}
50 
+
51 
+## Simple univariate imputation
52 
+.poplin_impute_simple <- function(x, type = c("halfmin", "median", "mean")) {
53 
+  type <- match.arg(type)
54 
+  switch(
55 
+    type,
56 
+    halfmin = .poplin_impute_halfmin(x),
57 
+    median = .poplin_impute_median(x),
58 
+    mean = .poplin_impute_mean(x)
59 
+  )
60 
+}
61 
+
17 62 
 .poplin_impute_halfmin <- function(x) {
18 63 
   out <- apply(x, 1, function(x) {
19 64 
     if (anyNA(x)) {
... ... 
@@ -46,33 +91,3 @@
46 91 
   })
47 92 
   t(out)
48 93 
 }
49 
-
50 
-.poplin_impute_knn <- function(x, by = c("feature", "sample"), ...) {
51 
-  if (!requireNamespace("VIM", quietly = TRUE)) {
52 
-    stop("Package 'VIM' is required. Please install and try again.")
53 
-  }
54 
-  by <- match.arg(by)
55 
-  if (by == "feature") {
56 
-    out <- VIM::kNN(x, ...)[, 1:ncol(x)]
57 
-    ## VIM package internally converts x as data.table, which drops rownames
58 
-    rownames(out) <- rownames(x)
59 
-  } else {
60 
-    out <- t(VIM::kNN(t(x), ...))[1:nrow(x), ]
61 
-    colnames(out) <- colnames(x)
62 
-  }
63 
-  as.matrix(out)
64 
-}
65 
-
66 
-.poplin_impute_pca <- function(x, ...) {
67 
-  if (!requireNamespace("pcaMethods", quietly = TRUE)) {
68 
-    stop("Package 'pcaMethods' is required. Please install and try again.")
69 
-  }
70 
-  t(pcaMethods::pca(t(x), method = "bpca", ...)@completeObs)
71 
-}
72 
-
73 
-.poplin_impute_randomforest <- function(x, ...) {
74 
-  if (!requireNamespace("missForest", quietly = TRUE)) {
75 
-    stop("Package 'missForest' is required. Please install and try again.")
76 
-  }
77 
-  t(missForest::missForest(t(x), ...)$ximp)
78 
-}
R/imputation-methods.R
History View file @ 492d166
... ... 
@@ -2,12 +2,34 @@
2 2 
 ##'
3 3 
 ##' Missing values are frequently found in metabolomics data. The \pkg{poplin}
4 4 
 ##' package provides a few options to handle them.
5 
-##'
5 
+##' [poplin_impute] is a wrapper for the following set of functions:
6 
+##' \describe{
7 
+##' \item{\code{\link{poplin_impute_knn}}:}{
8 
+##' k-nearest neighbor (KNN) imputation
9 
+##' }
10 
+##' \item{\code{\link{poplin_impute_pca}}:}{
11 
+##' principal component analysis (PCA) imputation
12 
+##' }
13 
+##' \item{\code{\link{poplin_impute_randomforest}}:}{
14 
+##' random forest imputation
15 
+##' }
16 
+##' \item{\code{\link{poplin_impute_simple}}:}{
17 
+##' simple univariate imputation (e.g., half-minimum, mean, median)
18 
+##' }
19 
+##' }
20 
+##' @param x A matrix or \linkS4class{poplin} object.
21 
+##' @param method A imputation method. Default is 'pqn'.
22 
+##' @param poplin_in Name of a data matrix to retrieve.
23 
+##' @param poplin_out Name of a data matrix to store.
24 
+##' @param ... Argument passed to a specific imputation method.
25 
+##' @return A matrix or \linkS4class{poplin} object of the same dimension as
26 
+##'   \code{x} containing the imputed intensities.
6 27 
 ##' @name poplin_impute
28 
+##' @family imputation methods
7 29 
 setMethod(
8 30 
   "poplin_impute",
9 31 
   "matrix",
10 
-  function(x, method, ...) {
32 
+  function(x, method = c("knn", "pca", "randomforest", "simple"), ...) {
11 33 
     .poplin_impute(x, method = method, ...)
12 34 
   }
13 35 
 )
... ... 
@@ -16,123 +38,233 @@ setMethod(
16 38 
 setMethod(
17 39 
   "poplin_impute",
18 40 
   "poplin",
19 
-  function(x, method, poplin_in, poplin_out, ...) {
41 
+  function(x, method = c("knn", "pca", "randomforest", "simple"),
42 
+           poplin_in, poplin_out, ...) {
20 43 
     m <- .verify_and_extract_input(x, poplin_in)
21 44 
     poplin_data(x, poplin_out) <- .poplin_impute(m, method = method, ...)
22 45 
     x
23 46 
   }
24 47 
 )
25 48
26 
-##' @rdname poplin_impute
49 
+##' K-nearest neighbor (KNN) imputation
50 
+##'
51 
+##' Apply k-nearest neighbor (KNN) imputation to a matrix or
52 
+##' \linkS4class{poplin} object. This is an interface to the [VIM::kNN] from the
53 
+##' \pkg{VIM} package. Since it is based on Gower's distance, standardization of
54 
+##' input data prior to KNN imputation would not affect the result.
55 
+##'
56 
+##' @references
57 
+##' Alexander Kowarik, Matthias Templ (2016). Imputation with the R Package VIM.
58 
+##' Journal of Statistical Software, 74(7), 1-16. doi:10.18637/jss.v074.i07
59 
+##'
60 
+##' Gower, J. C. (1971). A General Coefficient of Similarity and Some of Its
61 
+##' Properties. Biometrics, 27(4), 857–871. https://doi.org/10.2307/2528823
62 
+##'
63 
+##' @param x A matrix or \linkS4class{poplin} object.
64 
+##' @param poplin_in Name of a data matrix to retrieve.
65 
+##' @param poplin_out Name of a data matrix to store.
66 
+##' @param by Imputation by k-nearest features or by k-nearest samples.
67 
+##' @param ... Additional argument passed to [VIM::kNN].
68 
+##' @return A matrix or \linkS4class{poplin} object of the same dimension as
69 
+##'   \code{x} containing the imputed intensities.
70 
+##' @name poplin_impute_knn
71 
+##' @family imputation methods
27 72 
 setMethod(
28 73 
   "poplin_impute_knn",
29 74 
   "matrix",
30 
-  function(x, ...) {
31 
-    .poplin_impute_knn(x, ...)
75 
+  function(x, by = c("feature", "sample"),  ...) {
76 
+    .poplin_impute_knn(x, by = by, ...)
32 77 
   }
33 78 
 )
34 79
35 
-##' @rdname poplin_impute
80 
+##' @rdname poplin_impute_knn
36 81 
 setMethod(
37 82 
   "poplin_impute_knn",
38 83 
   "poplin",
39 
-  function(x, poplin_in, poplin_out, ...) {
84 
+  function(x, poplin_in, poplin_out, by = c("feature", "sample"), ...) {
40 85 
     .poplin_extract_and_assign(x, .poplin_impute_knn,
41 
-                               poplin_in, poplin_out, ...)
86 
+                               poplin_in, poplin_out,
87 
+                               by = by, ...)
42 88 
   }
43 89 
 )
44 90
45 
-##' @rdname poplin_impute
91 
+##' Random forest imputation
92 
+##'
93 
+##' Apply random forest imputation to a matrix or \linkS4class{poplin} object.
94 
+##' This is an interface to the [missForest::missForest] from the
95 
+##' \pkg{missForest} package. Since random forest is a tree-based method, it can
96 
+##' be performed with raw intensities - invariant to monotonic transformations
97 
+##' (However, statistical analysis could be affected because, for example,
98 
+##' log(mean(predicted values) != mean(log(predicted values))).
99 
+##'
100 
+##' @references
101 
+##'
102 
+##' Daniel J. Stekhoven (2013). missForest: Nonparametric Missing Value
103 
+##' Imputation using Random Forest. R package version 1.4.
104 
+##'
105 
+##' Stekhoven D. J., & Buehlmann, P. (2012). MissForest - non-parametric missing
106 
+##' value imputation for mixed-type data. Bioinformatics, 28(1), 112-118.
107 
+##'
108 
+##' @param x A matrix or \linkS4class{poplin} object.
109 
+##' @param poplin_in Name of a data matrix to retrieve.
110 
+##' @param poplin_out Name of a data matrix to store.
111 
+##' @param ... Additional argument passed to [missForest::missForest].
112 
+##' @return A matrix or \linkS4class{poplin} object of the same dimension as
113 
+##'   \code{x} containing the imputed intensities.
114 
+##' @name poplin_impute_randomforest
115 
+##' @family imputation methods
46 116 
 setMethod(
47 
-  "poplin_impute_halfmin",
117 
+  "poplin_impute_randomforest",
48 118 
   "matrix",
49 119 
   function(x, ...) {
50 
-    .poplin_impute_halfmin(x, ...)
120 
+    .poplin_impute_randomforest(x, ...)
51 121 
   }
52 122 
 )
53 123
54 
-##' @rdname poplin_impute
124 
+##' @rdname poplin_impute_randomforest
55 125 
 setMethod(
56 
-  "poplin_impute_halfmin",
126 
+  "poplin_impute_randomforest",
57 127 
   "poplin",
58 128 
   function(x, poplin_in, poplin_out, ...) {
59 
-    .poplin_extract_and_assign(x, .poplin_impute_halfmin,
129 
+    .poplin_extract_and_assign(x, .poplin_impute_randomforest,
60 130 
                                poplin_in, poplin_out, ...)
61 131 
   }
62 132 
 )
63 133
64 
-##' @rdname poplin_impute
134 
+##' Principal component analysis (PCA) imputation
135 
+##'
136 
+##' Apply PCA imputation to a matrix or \linkS4class{poplin} object. This is a
137 
+##' interface to the [pcaMethods::pca] from the \pkg{pcaMethods} package. Here,
138 
+##' features are interpreted as variables and samples as observations.
139 
+##' Pre-processing of input (centering, scaling) may be necessary. See the
140 
+##' documentation of [pcaMethods:pca] and [pcaMethods:prep]. Note that the PCA
141 
+##' imputation could yield negative feature values that need to be
142 
+##' post-processed.
143 
+##'
144 
+##' @references
145 
+##' Stacklies, W., Redestig, H., Scholz, M., Walther, D. and Selbig, J.
146 
+##' pcaMethods -- a Bioconductor package providing PCA methods for incomplete
147 
+##' data. Bioinformatics, 2007, 23, 1164-1167
148 
+##'
149 
+##' @param x A matrix or \linkS4class{poplin} object.
150 
+##' @param poplin_in Name of a data matrix to retrieve.
151 
+##' @param poplin_out Name of a data matrix to store.
152 
+##' @param type A method for performing PCA.
153 
+##' @param ... Additional argument passed to [pcaMethods::pca].
154 
+##' @return A matrix or \linkS4class{poplin} object of the same dimension as
155 
+##'   \code{x} containing the imputed intensities.
156 
+##' @name poplin_impute_pca
157 
+##' @family imputation methods
65 158 
 setMethod(
66 
-  "poplin_impute_median",
159 
+  "poplin_impute_pca",
67 160 
   "matrix",
68 
-  function(x, ...) {
69 
-    .poplin_impute_median(x, ...)
161 
+  function(x, type = c("bpca", "ppca", "nipals", "svdImpute"), ...) {
162 
+    .poplin_impute_pca(x, type = type, ...)
70 163 
   }
71 164 
 )
72 165
73 
-##' @rdname poplin_impute
166 
+##' @rdname poplin_impute_pca
74 167 
 setMethod(
75 
-  "poplin_impute_median",
168 
+  "poplin_impute_pca",
76 169 
   "poplin",
77 
-  function(x, poplin_in, poplin_out, ...) {
78 
-    .poplin_extract_and_assign(x, .poplin_impute_median,
79 
-                               poplin_in, poplin_out, ...)
170 
+  function(x, poplin_in, poplin_out,
171 
+           type = c("bpca", "ppca", "nipals", "svdImpute"), ...) {
172 
+    .poplin_extract_and_assign(x, .poplin_impute_pca,
173 
+                               poplin_in, poplin_out,
174 
+                               type = type, ...)
80 175 
   }
81 176 
 )
82 177
83 
-##' @rdname poplin_impute
178 
+##' Simple univariate imputation
179 
+##'
180 
+##' Apply univariate imputation to a matrix or \linkS4class{poplin} object. The
181 
+##' supported methods include
182 
+##' \itemize{
183 
+##' \item Half-minimum imputation: for each feature, missing values are replaced
184 
+##' with half the observed minimum.
185 
+##' \item Median imputation: for each feature, missing values are replaced with
186 
+##' the median of non-missing values.
187 
+##' \item Mean imputation: for each feature, missing values are replaced with
188 
+##' the mean of non-missing values.
189 
+##' }
190 
+##' @references
191 
+##' Wei, R., Wang, J., Su, M. et al. Missing Value Imputation Approach for Mass
192 
+##' Spectrometry-based Metabolomics Data. Sci Rep 8, 663 (2018).
193 
+##' https://doi.org/10.1038/s41598-017-19120-0
194 
+##'
195 
+##' @param x A matrix or \linkS4class{poplin} object.
196 
+##' @param poplin_in Name of a data matrix to retrieve.
197 
+##' @param poplin_out Name of a data matrix to store.
198 
+##' @param type A method for doing univariate imputation.
199 
+##' @return A matrix or \linkS4class{poplin} object of the same dimension as
200 
+##'   \code{x} containing the imputed intensities.
201 
+##' @name poplin_impute_simple
202 
+##' @family imputation methods
84 203 
 setMethod(
85 
-  "poplin_impute_mean",
204 
+  "poplin_impute_simple",
86 205 
   "matrix",
87 
-  function(x, ...) {
88 
-    .poplin_impute_mean(x, ...)
206 
+  function(x, type = c("halfmin", "median", "mean")) {
207 
+    .poplin_impute_simple(x, type = type)
89 208 
   }
90 209 
 )
91 210
92 
-##' @rdname poplin_impute
211 
+##' @rdname poplin_impute_simple
93 212 
 setMethod(
94 
-  "poplin_impute_mean",
213 
+  "poplin_impute_simple",
95 214 
   "poplin",
96 
-  function(x, poplin_in, poplin_out, ...) {
97 
-    .poplin_extract_and_assign(x, .poplin_impute_mean,
98 
-                               poplin_in, poplin_out, ...)
215 
+  function(x, poplin_in, poplin_out, type = c("halfmin", "median", "mean")) {
216 
+    .poplin_extract_and_assign(x, .poplin_impute_halfmin,
217 
+                               poplin_in, poplin_out, type = type)
99 218 
   }
100 219 
 )
101 220
102 
-##' @rdname poplin_impute
103 221 
 setMethod(
104 
-  "poplin_impute_pca",
222 
+  "poplin_impute_halfmin",
223 
+  "matrix",
224 
+  function(x) {
225 
+    .poplin_impute_halfmin(x)
226 
+  }
227 
+)
228 
+
229 
+setMethod(
230 
+  "poplin_impute_halfmin",
231 
+  "poplin",
232 
+  function(x, poplin_in, poplin_out) {
233 
+    .poplin_extract_and_assign(x, .poplin_impute_halfmin,
234 
+                               poplin_in, poplin_out)
235 
+  }
236 
+)
237 
+
238 
+setMethod(
239 
+  "poplin_impute_median",
105 240 
   "matrix",
106 241 
   function(x, ...) {
107 
-    .poplin_impute_pca(x, ...)
242 
+    .poplin_impute_median(x, ...)
108 243 
   }
109 244 
 )
110 245
111 
-##' @rdname poplin_impute
112 246 
 setMethod(
113 
-  "poplin_impute_pca",
247 
+  "poplin_impute_median",
114 248 
   "poplin",
115 249 
   function(x, poplin_in, poplin_out, ...) {
116 
-    .poplin_extract_and_assign(x, .poplin_impute_pca,
250 
+    .poplin_extract_and_assign(x, .poplin_impute_median,
117 251 
                                poplin_in, poplin_out, ...)
118 252 
   }
119 253 
 )
120 254
121 
-##' @rdname poplin_impute
122 255 
 setMethod(
123 
-  "poplin_impute_randomforest",
256 
+  "poplin_impute_mean",
124 257 
   "matrix",
125 258 
   function(x, ...) {
126 
-    .poplin_impute_randomforest(x, ...)
259 
+    .poplin_impute_mean(x, ...)
127 260 
   }
128 261 
 )
129 262
130 
-##' @rdname poplin_impute
131 263 
 setMethod(
132 
-  "poplin_impute_randomforest",
264 
+  "poplin_impute_mean",
133 265 
   "poplin",
134 266 
   function(x, poplin_in, poplin_out, ...) {
135 
-    .poplin_extract_and_assign(x, .poplin_impute_randomforest,
267 
+    .poplin_extract_and_assign(x, .poplin_impute_mean,
136 268 
                                poplin_in, poplin_out, ...)
137 269 
   }
138 270 
 )
R/normalization-functions.R
History View file @ 492d166
... ... 
@@ -394,7 +394,7 @@
394 394 
     range = .poplin_normalize_range(x),
395 395 
     pareto = .poplin_normalize_pareto(x),
396 396 
     vast = .poplin_normalize_vast(x),
397 
-    level = .poplin_normalize_level(x),
397 
+    level = .poplin_normalize_level(x)
398 398 
   )
399 399 
 }
400 400
R/normalization-methods.R
History View file @ 492d166
... ... 
@@ -63,13 +63,13 @@ setMethod(
63 63
64 64 
 ##' Probabilistic quotient normalization (PQN)
65 65 
 ##'
66 
-##' Apply probabilistic quotient normalization to a matrix or
66 
+##' Apply probabilistic quotient normalization (PQN) to a matrix or
67 67 
 ##' \linkS4class{poplin} object. For the calculation of quotients, a reference
68 68 
 ##' spectrum needs to be obtained from a mean or median spectrum based on all
69 69 
 ##' spectra of the study or a subset of the study. Feature intensities are
70 70 
 ##' normalized by the median of quotients. See Dieterle et al. (2006) for
71 71 
 ##' details.
72 
-##'
72 
+##'
73 73 
 ##' @param x A matrix or \linkS4class{poplin} object.
74 74 
 ##' @param poplin_in Name of a data matrix to retrieve.
75 75 
 ##' @param poplin_out Name of a data matrix to store.
... ... 
@@ -116,7 +116,7 @@ setMethod(
116 116 
 ##' Apply sum normalization to a matrix or \linkS4class{poplin} object. For each
117 117 
 ##' sample, feature intensities are divided by its Total Ion Current (TIC),
118 118 
 ##' i.e., every feature is divided by the sum of all intensity values.
119 
-##'
119 
+##'
120 120 
 ##' @param x A matrix or \linkS4class{poplin} object.
121 121 
 ##' @param poplin_in Name of a data matrix to retrieve.
122 122 
 ##' @param poplin_out Name of a data matrix to store.
... ... 
@@ -153,7 +153,7 @@ setMethod(
153 153 
 ##' Apply mean normalization to a matrix or \linkS4class{poplin} object. For
154 154 
 ##' each sample, feature intensities are divided by its mean. The mean of
155 155 
 ##' intensity values for individual samples will be one as a result.
156 
-##'
156 
+##'
157 157 
 ##' @param x A matrix or \linkS4class{poplin} object.
158 158 
 ##' @param poplin_in Name of a data matrix to retrieve.
159 159 
 ##' @param poplin_out Name of a data matrix to store.
... ... 
@@ -190,7 +190,7 @@ setMethod(
190 190 
 ##' Apply median normalization to a matrix or \linkS4class{poplin} object. For
191 191 
 ##' each sample, feature intensities are divided by its median. The median of
192 192 
 ##' intensity values for individual samples will be one as a result.
193 
-##'
193 
+##'
194 194 
 ##' @param x A matrix or \linkS4class{poplin} object.
195 195 
 ##' @param poplin_in Name of a data matrix to retrieve.
196 196 
 ##' @param poplin_out Name of a data matrix to store.
... ... 
@@ -228,7 +228,7 @@ setMethod(
228 228 
 ##' \linkS4class{poplin} object. For each sample, feature intensities are scaled
229 229 
 ##' by its MAD. The MAD of intensity values for individual samples will be one
230 230 
 ##' as a result.
231 
-##'
231 
+##'
232 232 
 ##' @param x A matrix or \linkS4class{poplin} object.
233 233 
 ##' @param poplin_in Name of a data matrix to retrieve.
234 234 
 ##' @param poplin_out Name of a data matrix to store.
... ... 
@@ -347,7 +347,7 @@ setMethod(
347 347 
 ##' from the \pkg{vsn} package (see [vsn::vsn2] for help). The vsn produces
348 348 
 ##' normalized intensities based on a glog (generalized logarithm) scale to base
349 349 
 ##' 2. See Huber et al. (2002) for details.
350 
-##'
350 
+##'
351 351 
 ##' @references
352 352 
 ##' Huber W, von Heydebreck A, Sültmann H, Poustka A, Vingron M. Variance
353 353 
 ##' stabilization applied to microarray data calibration and to the
... ... 
@@ -384,7 +384,7 @@ setMethod(
384 384 
 )
385 385
386 386 
 ##' Feature-based scaling
387 
-##'
387 
+##'
388 388 
 ##' Apply feature-based scaling to a matrix or \linkS4class{poplin} object. The
389 389 
 ##' supported methods include
390 390 
 ##' \itemize{
man/poplin_impute.Rd
History View file @ 492d166
... ... 
@@ -3,49 +3,57 @@
3 3 
 \name{poplin_impute}
4 4 
 \alias{poplin_impute}
5 5 
 \alias{poplin_impute,poplin-method}
6 
-\alias{poplin_impute_knn,matrix-method}
7 
-\alias{poplin_impute_knn,poplin-method}
8 
-\alias{poplin_impute_halfmin,matrix-method}
9 
-\alias{poplin_impute_halfmin,poplin-method}
10 
-\alias{poplin_impute_median,matrix-method}
11 
-\alias{poplin_impute_median,poplin-method}
12 
-\alias{poplin_impute_mean,matrix-method}
13 
-\alias{poplin_impute_mean,poplin-method}
14 
-\alias{poplin_impute_pca,matrix-method}
15 
-\alias{poplin_impute_pca,poplin-method}
16 
-\alias{poplin_impute_randomforest,matrix-method}
17 
-\alias{poplin_impute_randomforest,poplin-method}
18 6 
 \title{Imputation methods}
19 7 
 \usage{
20 
-\S4method{poplin_impute}{matrix}(x, method, ...)
21 
-
22 
-\S4method{poplin_impute}{poplin}(x, method, poplin_in, poplin_out, ...)
23 
-
24 
-\S4method{poplin_impute_knn}{matrix}(x, ...)
25 
-
26 
-\S4method{poplin_impute_knn}{poplin}(x, poplin_in, poplin_out, ...)
27 
-
28 
-\S4method{poplin_impute_halfmin}{matrix}(x, ...)
29 
-
30 
-\S4method{poplin_impute_halfmin}{poplin}(x, poplin_in, poplin_out, ...)
31 
-
32 
-\S4method{poplin_impute_median}{matrix}(x, ...)
33 
-
34 
-\S4method{poplin_impute_median}{poplin}(x, poplin_in, poplin_out, ...)
35 
-
36 
-\S4method{poplin_impute_mean}{matrix}(x, ...)
37 
-
38 
-\S4method{poplin_impute_mean}{poplin}(x, poplin_in, poplin_out, ...)
8 
+\S4method{poplin_impute}{matrix}(x, method = c("knn", "pca", "randomforest", "simple"), ...)
9 
+
10 
+\S4method{poplin_impute}{poplin}(
11 
+  x,
12 
+  method = c("knn", "pca", "randomforest", "simple"),
13 
+  poplin_in,
14 
+  poplin_out,
15 
+  ...
16 
+)
17 
+}
18 
+\arguments{
19 
+\item{x}{A matrix or \linkS4class{poplin} object.}
39 20
40 
-\S4method{poplin_impute_pca}{matrix}(x, ...)
21 
+\item{method}{A imputation method. Default is 'pqn'.}
41 22
42 
-\S4method{poplin_impute_pca}{poplin}(x, poplin_in, poplin_out, ...)
23 
+\item{...}{Argument passed to a specific imputation method.}
43 24
44 
-\S4method{poplin_impute_randomforest}{matrix}(x, ...)
25 
+\item{poplin_in}{Name of a data matrix to retrieve.}
45 26
46 
-\S4method{poplin_impute_randomforest}{poplin}(x, poplin_in, poplin_out, ...)
27 
+\item{poplin_out}{Name of a data matrix to store.}
28 
+}
29 
+\value{
30 
+A matrix or \linkS4class{poplin} object of the same dimension as
31 
+\code{x} containing the imputed intensities.
47 32 
 }
48 33 
 \description{
49 34 
 Missing values are frequently found in metabolomics data. The \pkg{poplin}
50 35 
 package provides a few options to handle them.
36 
+\link{poplin_impute} is a wrapper for the following set of functions:
37 
+\describe{
38 
+\item{\code{\link{poplin_impute_knn}}:}{
39 
+k-nearest neighbor (KNN) imputation
40 
+}
41 
+\item{\code{\link{poplin_impute_pca}}:}{
42 
+principal component analysis (PCA) imputation
43 
+}
44 
+\item{\code{\link{poplin_impute_randomforest}}:}{
45 
+random forest imputation
46 
+}
47 
+\item{\code{\link{poplin_impute_simple}}:}{
48 
+simple univariate imputation (e.g., half-minimum, mean, median)
49 
+}
50 
+}
51 
+}
52 
+\seealso{
53 
+Other imputation methods:
54 
+\code{\link{poplin_impute_knn}()},
55 
+\code{\link{poplin_impute_pca}()},
56 
+\code{\link{poplin_impute_randomforest}()},
57 
+\code{\link{poplin_impute_simple}()}
51 58 
 }
59 
+\concept{imputation methods}
man/poplin_impute_knn.Rd
History View file @ 492d166
52 60 
new file mode 100644
... ... 
@@ -0,0 +1,47 @@
1 
+% Generated by roxygen2: do not edit by hand
2 
+% Please edit documentation in R/imputation-methods.R
3 
+\name{poplin_impute_knn}
4 
+\alias{poplin_impute_knn}
5 
+\alias{poplin_impute_knn,poplin-method}
6 
+\title{K-nearest neighbor (KNN) imputation}
7 
+\usage{
8 
+\S4method{poplin_impute_knn}{matrix}(x, by = c("feature", "sample"), ...)
9 
+
10 
+\S4method{poplin_impute_knn}{poplin}(x, poplin_in, poplin_out, by = c("feature", "sample"), ...)
11 
+}
12 
+\arguments{
13 
+\item{x}{A matrix or \linkS4class{poplin} object.}
14 
+
15 
+\item{by}{Imputation by k-nearest features or by k-nearest samples.}
16 
+
17 
+\item{...}{Additional argument passed to \link[VIM:kNN]{VIM::kNN}.}
18 
+
19 
+\item{poplin_in}{Name of a data matrix to retrieve.}
20 
+
21 
+\item{poplin_out}{Name of a data matrix to store.}
22 
+}
23 
+\value{
24 
+A matrix or \linkS4class{poplin} object of the same dimension as
25 
+\code{x} containing the imputed intensities.
26 
+}
27 
+\description{
28 
+Apply k-nearest neighbor (KNN) imputation to a matrix or
29 
+\linkS4class{poplin} object. This is an interface to the \link[VIM:kNN]{VIM::kNN} from the
30 
+\pkg{VIM} package. Since it is based on Gower's distance, standardization of
31 
+input data prior to KNN imputation would not affect the result.
32 
+}
33 
+\references{
34 
+Alexander Kowarik, Matthias Templ (2016). Imputation with the R Package VIM.
35 
+Journal of Statistical Software, 74(7), 1-16. doi:10.18637/jss.v074.i07
36 
+
37 
+Gower, J. C. (1971). A General Coefficient of Similarity and Some of Its
38 
+Properties. Biometrics, 27(4), 857–871. https://doi.org/10.2307/2528823
39 
+}
40 
+\seealso{
41 
+Other imputation methods:
42 
+\code{\link{poplin_impute_pca}()},
43 
+\code{\link{poplin_impute_randomforest}()},
44 
+\code{\link{poplin_impute_simple}()},
45 
+\code{\link{poplin_impute}()}
46 
+}
47 
+\concept{imputation methods}
man/poplin_impute_pca.Rd
History View file @ 492d166
0 48 
new file mode 100644
... ... 
@@ -0,0 +1,54 @@
1 
+% Generated by roxygen2: do not edit by hand
2 
+% Please edit documentation in R/imputation-methods.R
3 
+\name{poplin_impute_pca}
4 
+\alias{poplin_impute_pca}
5 
+\alias{poplin_impute_pca,poplin-method}
6 
+\title{Principal component analysis (PCA) imputation}
7 
+\usage{
8 
+\S4method{poplin_impute_pca}{matrix}(x, type = c("bpca", "ppca", "nipals", "svdImpute"), ...)
9 
+
10 
+\S4method{poplin_impute_pca}{poplin}(
11 
+  x,
12 
+  poplin_in,
13 
+  poplin_out,
14 
+  type = c("bpca", "ppca", "nipals", "svdImpute"),
15 
+  ...
16 
+)
17 
+}
18 
+\arguments{
19 
+\item{x}{A matrix or \linkS4class{poplin} object.}
20 
+
21 
+\item{type}{A method for performing PCA.}
22 
+
23 
+\item{...}{Additional argument passed to \link[pcaMethods:pca]{pcaMethods::pca}.}
24 
+
25 
+\item{poplin_in}{Name of a data matrix to retrieve.}
26 
+
27 
+\item{poplin_out}{Name of a data matrix to store.}
28 
+}
29 
+\value{
30 
+A matrix or \linkS4class{poplin} object of the same dimension as
31 
+\code{x} containing the imputed intensities.
32 
+}
33 
+\description{
34 
+Apply PCA imputation to a matrix or \linkS4class{poplin} object. This is a
35 
+interface to the \link[pcaMethods:pca]{pcaMethods::pca} from the \pkg{pcaMethods} package. Here,
36 
+features are interpreted as variables and samples as observations.
37 
+Pre-processing of input (centering, scaling) may be necessary. See the
38 
+documentation of \link{pcaMethods:pca} and \link{pcaMethods:prep}. Note that the PCA
39 
+imputation could yield negative feature values that need to be
40 
+post-processed.
41 
+}
42 
+\references{
43 
+Stacklies, W., Redestig, H., Scholz, M., Walther, D. and Selbig, J.
44 
+pcaMethods -- a Bioconductor package providing PCA methods for incomplete
45 
+data. Bioinformatics, 2007, 23, 1164-1167
46 
+}
47 
+\seealso{
48 
+Other imputation methods:
49 
+\code{\link{poplin_impute_knn}()},
50 
+\code{\link{poplin_impute_randomforest}()},
51 
+\code{\link{poplin_impute_simple}()},
52 
+\code{\link{poplin_impute}()}
53 
+}
54 
+\concept{imputation methods}
man/poplin_impute_randomforest.Rd
History View file @ 492d166
0 55 
new file mode 100644
... ... 
@@ -0,0 +1,47 @@
1 
+% Generated by roxygen2: do not edit by hand
2 
+% Please edit documentation in R/imputation-methods.R
3 
+\name{poplin_impute_randomforest}
4 
+\alias{poplin_impute_randomforest}
5 
+\alias{poplin_impute_randomforest,poplin-method}
6 
+\title{Random forest imputation}
7 
+\usage{
8 
+\S4method{poplin_impute_randomforest}{matrix}(x, ...)
9 
+
10 
+\S4method{poplin_impute_randomforest}{poplin}(x, poplin_in, poplin_out, ...)
11 
+}
12 
+\arguments{
13 
+\item{x}{A matrix or \linkS4class{poplin} object.}
14 
+
15 
+\item{...}{Additional argument passed to \link[missForest:missForest]{missForest::missForest}.}
16 
+
17 
+\item{poplin_in}{Name of a data matrix to retrieve.}
18 
+
19 
+\item{poplin_out}{Name of a data matrix to store.}
20 
+}
21 
+\value{
22 
+A matrix or \linkS4class{poplin} object of the same dimension as
23 
+\code{x} containing the imputed intensities.
24 
+}
25 
+\description{
26 
+Apply random forest imputation to a matrix or \linkS4class{poplin} object.
27 
+This is an interface to the \link[missForest:missForest]{missForest::missForest} from the
28 
+\pkg{missForest} package. Since random forest is a tree-based method, it can
29 
+be performed with raw intensities - invariant to monotonic transformations
30 
+(However, statistical analysis could be affected because, for example,
31 
+log(mean(predicted values) != mean(log(predicted values))).
32 
+}
33 
+\references{
34 
+Daniel J. Stekhoven (2013). missForest: Nonparametric Missing Value
35 
+Imputation using Random Forest. R package version 1.4.
36 
+
37 
+Stekhoven D. J., & Buehlmann, P. (2012). MissForest - non-parametric missing
38 
+value imputation for mixed-type data. Bioinformatics, 28(1), 112-118.
39 
+}
40 
+\seealso{
41 
+Other imputation methods:
42 
+\code{\link{poplin_impute_knn}()},
43 
+\code{\link{poplin_impute_pca}()},
44 
+\code{\link{poplin_impute_simple}()},
45 
+\code{\link{poplin_impute}()}
46 
+}
47 
+\concept{imputation methods}
man/poplin_impute_simple.Rd
History View file @ 492d166
0 48 
new file mode 100644
... ... 
@@ -0,0 +1,54 @@
1 
+% Generated by roxygen2: do not edit by hand
2 
+% Please edit documentation in R/imputation-methods.R
3 
+\name{poplin_impute_simple}
4 
+\alias{poplin_impute_simple}
5 
+\alias{poplin_impute_simple,poplin-method}
6 
+\title{Simple univariate imputation}
7 
+\usage{
8 
+\S4method{poplin_impute_simple}{matrix}(x, type = c("halfmin", "median", "mean"))
9 
+
10 
+\S4method{poplin_impute_simple}{poplin}(
11 
+  x,
12 
+  poplin_in,
13 
+  poplin_out,
14 
+  type = c("halfmin", "median", "mean")
15 
+)
16 
+}
17 
+\arguments{
18 
+\item{x}{A matrix or \linkS4class{poplin} object.}
19 
+
20 
+\item{type}{A method for doing univariate imputation.}
21 
+
22 
+\item{poplin_in}{Name of a data matrix to retrieve.}
23 
+
24 
+\item{poplin_out}{Name of a data matrix to store.}
25 
+}
26 
+\value{
27 
+A matrix or \linkS4class{poplin} object of the same dimension as
28 
+\code{x} containing the imputed intensities.
29 
+}
30 
+\description{
31 
+Apply univariate imputation to a matrix or \linkS4class{poplin} object. The
32 
+supported methods include
33 
+\itemize{
34 
+\item Half-minimum imputation: for each feature, missing values are replaced
35 
+with half the observed minimum.
36 
+\item Median imputation: for each feature, missing values are replaced with
37 
+the median of non-missing values.
38 
+\item Mean imputation: for each feature, missing values are replaced with
39 
+the mean of non-missing values.
40 
+}
41 
+}
42 
+\references{
43 
+Wei, R., Wang, J., Su, M. et al. Missing Value Imputation Approach for Mass
44 
+Spectrometry-based Metabolomics Data. Sci Rep 8, 663 (2018).
45 
+https://doi.org/10.1038/s41598-017-19120-0
46 
+}
47 
+\seealso{
48 
+Other imputation methods:
49 
+\code{\link{poplin_impute_knn}()},
50 
+\code{\link{poplin_impute_pca}()},
51 
+\code{\link{poplin_impute_randomforest}()},
52 
+\code{\link{poplin_impute}()}
53 
+}
54 
+\concept{imputation methods}
man/poplin_normalize_pqn.Rd
History View file @ 492d166
... ... 
@@ -43,7 +43,7 @@ A matrix or \linkS4class{poplin} object of the same dimension as
43 43 
 \code{x} containing the normalized intensities.
44 44 
 }
45 45 
 \description{
46 
-Apply probabilistic quotient normalization to a matrix or
46 
+Apply probabilistic quotient normalization (PQN) to a matrix or
47 47 
 \linkS4class{poplin} object. For the calculation of quotients, a reference
48 48 
 spectrum needs to be obtained from a mean or median spectrum based on all
49 49 
 spectra of the study or a subset of the study. Feature intensities are