plot: Plot Method for (O)PLS(-DA)

In ropls: PCA, PLS(-DA) and OPLS(-DA) for multivariate analysis and feature selection of omics data

Description

This function plots values based upon a model trained by opls.

This function plots values based upon a model trained by opls.

Usage

## S4 method for signature 'oplsMultiDataSet,ANY'
plot(
  x,
  y,
  fig.pdfC = c("none", "interactive", "myfile.pdf")[2],
  info.txtC = c("none", "interactive", "myfile.txt")[2],
  ...
)

## S4 method for signature 'opls,ANY'
plot(
  x,
  y,
  typeVc = c("correlation", "outlier", "overview", "permutation", "predict-train",
    "predict-test", "summary", "x-loading", "x-score", "x-variance", "xy-score",
    "xy-weight")[7],
  parAsColFcVn = NA,
  parCexN = 0.8,
  parCompVi = c(1, 2),
  parEllipsesL = NA,
  parLabVc = NA,
  parPaletteVc = NA,
  parTitleL = TRUE,
  parCexMetricN = NA,
  plotPhenoDataC = NA,
  plotSubC = NA,
  fig.pdfC = c("none", "interactive", "myfile.pdf")[2],
  info.txtC = c("none", "interactive", "myfile.txt")[2],
  file.pdfC = NULL,
  .sinkC = NULL,
  ...
)

Arguments

x	An S4 object of class opls or oplsMultiDataSet, created by the opls function.
y	Currently not used
fig.pdfC	Character: File name with '.pdf' extension for the figure; if 'interactive' (default), figures will be displayed interactively; if 'none', no figure will be generated
info.txtC	Character: File name with '.txt' extension for the printed results (call to sink()'); if 'interactive' (default), messages will be printed on the screen; if 'none', no verbose will be generated
...	Currently not used.
typeVc	Character vector: the following plots are available: 'correlation': Variable correlations with the components, 'outlier': Observation diagnostics (score and orthogonal distances), 'overview': Model overview showing R2Ycum and Q2cum (or 'Variance explained' for PCA), 'permutation': Scatterplot of R2Y and Q2Y actual and simulated models after random permutation of response values; 'predict-train' and 'predict-test': Predicted vs Actual Y for reference and test sets (only if Y has a single column), 'summary' [default]: 4-plot summary showing permutation, overview, outlier, and x-score together, 'x-variance': Spread of raw variables corresp. with min, median, and max variances, 'x-loading': X-loadings (the 6 of variables most contributing to loadings are colored in red to facilitate interpretation), 'x-score': X-Scores, 'xy-score': XY-Scores, 'xy-weight': XY-Weights
parAsColFcVn	Optional factor character or numeric vector to be converted into colors for the score plot; default is NA [ie colors will be converted from 'y' in case of (O)PLS(-DA) or will be 'black' for PCA]
parCexN	Numeric: amount by which plotting text should be magnified relative to the default
parCompVi	Integer vector of length 2: indices of the two components to be displayed on the score plot (first two components by default)
parEllipsesL	Should the Mahalanobis ellipses be drawn? If 'NA' [default], ellipses are drawn when either a character parAsColVcn is provided (PCA case), or when 'y' is a character factor ((O)PLS-DA cases).
parLabVc	Optional character vector for the labels of observations on the plot; default is NA [ie row names of 'x', if available, or indices of 'x', otherwise, will be used]
parPaletteVc	Optional character vector of colors to be used in the plots
parTitleL	Should the titles of the plots be printed on the graphics (default = TRUE); It may be convenient to set this argument to FALSE when the user wishes to add specific titles a posteriori
parCexMetricN	Numeric: magnification of the metrics at the bottom of score plot (default -NA- is 1 in 1x1 and 0.7 in 2x2 display)
plotPhenoDataC	Character: if x was generated from an ExpressionSet (i.e. if the 'eset' slot from x is not NULL), the name of the pData(x) column to be used for coloring can be specified here (instead of 'parAsColFcVn')
plotSubC	Character: Graphic subtitle
file.pdfC	Character: deprecated; use the 'fig.pdfC' argument instead
.sinkC	Character: deprecated; use the 'info.txtC' argument instead

Examples

# Loading the 'NCI60_4arrays' from the 'omicade4' package
data("NCI60_4arrays", package = "omicade4")
# Selecting two of the four datasets
setNamesVc <- c("agilent", "hgu95")
# Creating the MultiDataSet instance
nciMset <- MultiDataSet::createMultiDataSet()
# Adding the two datasets as ExpressionSet instances
for (setC in setNamesVc) {
  # Getting the data
  exprMN <- as.matrix(NCI60_4arrays[[setC]])
  pdataDF <- data.frame(row.names = colnames(exprMN),
                        cancer = substr(colnames(exprMN), 1, 2),
                        stringsAsFactors = FALSE)
  fdataDF <- data.frame(row.names = rownames(exprMN),
                        name = rownames(exprMN),
                        stringsAsFactors = FALSE)
  # Building the ExpressionSet
  eset <- Biobase::ExpressionSet(assayData = exprMN,
                                 phenoData = new("AnnotatedDataFrame",
                                                 data = pdataDF),
                                 featureData = new("AnnotatedDataFrame",
                                                   data = fdataDF),
                                 experimentData = new("MIAME",
                                                      title = setC))
  # Adding to the MultiDataSet
  nciMset <- MultiDataSet::add_eset(nciMset, eset, dataset.type = setC,
                                    GRanges = NA, warnings = FALSE)
}
# Summary of the MultiDataSet
nciMset
# Principal Component Analysis of each data set
nciPca <- ropls::opls(nciMset)
# Coloring the Score plot according to cancer types
ropls::plot(nciPca, y = "cancer", typeVc = "x-score")
# Restricting to the 'ME' and 'LE' cancer types
sampleNamesVc <- Biobase::sampleNames(nciMset[["agilent"]])
cancerTypeVc <- Biobase::pData(nciMset[["agilent"]])[, "cancer"]
nciMset <- nciMset[sampleNamesVc[cancerTypeVc %in% c("ME", "LE")], ]
# Building PLS-DA models for the cancer type
nciPlsda <- ropls::opls(nciMset, "cancer", predI = 2)

data(sacurine)
attach(sacurine)

for(typeC in c("correlation", "outlier", "overview",
               "permutation", "predict-train","predict-test",
               "summary", "x-loading", "x-score", "x-variance",
               "xy-score", "xy-weight")) {

    print(typeC)

    if(grepl("predict", typeC))
        subset <- "odd"
    else
        subset <- NULL

    plsModel <- opls(dataMatrix, sampleMetadata[, "gender"],
                     predI = ifelse(typeC != "xy-weight", 1, 2),
                     orthoI = ifelse(typeC != "xy-weight", 1, 0),
                     permI = ifelse(typeC == "permutation", 10, 0),
                     subset = subset,
                     info.txtC = "none",
                     fig.pdfC = "none")

    plot(plsModel, typeVc = typeC)

}

sacPlsda <- opls(dataMatrix, sampleMetadata[, "gender"])
plot(sacPlsda, parPaletteVc = c("green4", "magenta"))

#### Application to an ExpressionSet

sacSet <- Biobase::ExpressionSet(assayData = t(dataMatrix), 
                                 phenoData = new("AnnotatedDataFrame", 
                                                 data = sampleMetadata), 
                                 featureData = new("AnnotatedDataFrame", 
                                                   data = variableMetadata),
                                 experimentData = new("MIAME", 
                                                      title = "sacurine"))
                                                      
sacPlsda <- opls(sacSet, "gender")
plot(sacPlsda, "gender", typeVc = "x-score")

detach(sacurine)

ropls documentation built on Nov. 8, 2020, 7:46 p.m.