esetLda {esetVis} | R Documentation |
esetLda
reduces the dimension of the data contained in the eSet via a linear discriminant analysis
on the specified grouping variable with the lda
function and plot the subsequent biplot,
possibly with sample annotation and gene annotation contained in the eSet.
esetLda( eset, ldaVar, psids = 1:nrow(eset), dim = c(1, 2), colorVar = character(), color = if (length(colorVar) == 0) "black" else character(), shapeVar = character(), shape = if (length(shapeVar) == 0) 15 else numeric(), sizeVar = character(), size = if (length(sizeVar) == 0) { ifelse(typePlot[1] == "interactive" && packageInteractivity[1] == "rbokeh", 5, 2.5) } else { numeric() }, sizeRange = numeric(), alphaVar = character(), alpha = if (length(alphaVar) == 0) 1 else numeric(), alphaRange = numeric(), title = "", symmetryAxes = c("combine", "separate", "none"), packageTextLabel = c("ggrepel", "ggplot2"), cloudGenes = TRUE, cloudGenesColor = "black", cloudGenesNBins = sqrt(length(psids)), cloudGenesIncludeLegend = FALSE, cloudGenesTitleLegend = "nGenes", topGenes = 10, topGenesCex = 2.5, topGenesVar = character(), topGenesJust = c(0.5, 0.5), topGenesColor = "black", topSamples = 10, topSamplesCex = 2.5, topSamplesVar = character(), topSamplesJust = c(0.5, 0.5), topSamplesColor = "black", geneSets = list(), geneSetsVar = character(), geneSetsMaxNChar = numeric(), topGeneSets = 10, topGeneSetsCex = 2.5, topGeneSetsJust = c(0.5, 0.5), topGeneSetsColor = "black", includeLegend = TRUE, includeLineOrigin = TRUE, typePlot = c("static", "interactive"), packageInteractivity = c("rbokeh", "ggvis"), figInteractiveSize = c(600, 400), ggvisAdjustLegend = TRUE, interactiveTooltip = TRUE, interactiveTooltipExtraVars = character(), returnAnalysis = FALSE, returnEsetPlot = FALSE )
eset |
expressionSet (or SummarizedExperiment) object with data |
ldaVar |
name of variable (in varLabels of the |
psids |
featureNames of genes to include in the plot, all by default |
dim |
dimensions of the analysis to represent, first two dimensions by default |
colorVar |
name of variable (in varLabels of the |
color |
character or factor with specified color(s) for the points,
replicated if needed. This is used only if |
shapeVar |
name of variable (in varLabels of the |
shape |
character or factor with specified shape(s) (pch) for the points,
replicated if needed. This is used only if |
sizeVar |
name of variable (in varLabels of the |
size |
character or factor with specified size(s) (cex) for the points,
replicated if needed.
This is used only if |
sizeRange |
size (cex) range used in the plot, possible only
if the |
alphaVar |
name of variable (in varLabels of the |
alpha |
character or factor with specified transparency(s) for the points,
replicated if needed. This is used only if |
alphaRange |
transparency (alpha) range used in the plot,
possible only if the |
title |
plot title, ” by default |
symmetryAxes |
set symmetry for axes, either:
|
packageTextLabel |
package used to label the outlying genes/samples/gene sets,
either |
cloudGenes |
logical, if TRUE (by default), include the cloud of genes in the plot |
cloudGenesColor |
if |
cloudGenesNBins |
number of bins to used for the clouds of genes, by default the square root of the number of genes |
cloudGenesIncludeLegend |
logical, if TRUE (FALSE by default) include the legend for the cloud of genes (in the top position if multiple legends) |
cloudGenesTitleLegend |
string with title for the legend for the cloud of genes 'nGenes' by default |
topGenes |
numeric indicating which percentile (if <1) or number (if >=1) of genes most distant to the origin of the plot to annotate, by default: 10 genes are selected If no genes should be annotated, set this parameter to 0 Currently only available for static plot. |
topGenesCex |
cex for gene annotation (used when |
topGenesVar |
variable of the featureData used to label the genes,
by default: empty, the featureNames are used for labelling (used when |
topGenesJust |
text justification for the genes
(used when |
topGenesColor |
text color for the genes
(used when |
topSamples |
numeric indicating which percentile (if <1) or number (if >=1) of samples most distant to the origin of the plot to annotate, by default: 10 samples are selected If no samples should be annotated, set this parameter to 0. Currently available for static plot. |
topSamplesCex |
cex for sample annotation (used when |
topSamplesVar |
variable of the phenoData used to label the samples,
by default: empty, the sampleNames are used for labelling
(used when |
topSamplesJust |
text justification for the samples
(used when |
topSamplesColor |
text color for the samples
(used when |
geneSets |
list of gene sets/pathways, each containing
identifiers of genes contained in the set.
E.g. pathways from Gene Ontology databases output from the
|
geneSetsVar |
variable of the featureData used to
match the genes contained in geneSets,
most probably ENTREZID, if not specified the featureNames
of the eSet are used
Only used when |
geneSetsMaxNChar |
maximum number of characters for pathway names,
by default keep entire names
Only used when |
topGeneSets |
numeric indicating which percentile (if <=1) or number (if >1) of gene sets
most distant to the origin of the plot to annotate, by default: 10 gene sets are selected
If no gene sets should be annotated, set this parameter to 0.
Currently available for static plot.
Only used when |
topGeneSetsCex |
cex for gene sets annotation
Only used when |
topGeneSetsJust |
text justification for the gene sets
by default: c(0.5, 0.5) so centered
Only used when |
topGeneSetsColor |
color for the gene sets
(used when |
includeLegend |
logical if TRUE (by default) include a legend, otherwise not |
includeLineOrigin |
if TRUE (by default) include vertical line at x = 0 and horizontal line at y = 0 |
typePlot |
type of the plot returned, either 'static' (static) or interactive' (potentially interactive) |
packageInteractivity |
if |
figInteractiveSize |
vector containing the size of the interactive plot,
as [width, height]
by default: c(600, 400). This is passed to the
|
ggvisAdjustLegend |
logical, if TRUE (by default) adjust the legends in |
interactiveTooltip |
logical, if TRUE, add hoover functionality showing sample annotation (variables used in the plot) in the plot |
interactiveTooltipExtraVars |
name of extra variable(s)
(in varLabels of the |
returnAnalysis |
logical, if TRUE (FALSE by default), return also the output of the analysis, and the outlying samples in the topElements element if any, otherwise only the plot object |
returnEsetPlot |
logical, if TRUE return also the esetPlot object |
if returnAnalysis
is TRUE, return a list:
analysis: output of the spectral map analysis, whose parameters
can be given as input to the esetPlotWrapper
function
dataPlotSamples: coordinates of the samples
dataPlotGenes: coordinates of the genes
esetUsed: expressionSet used in the plot
topElements: list with top outlying elements if any, possibly genes, samples and gene sets
plot: the plot output
otherwise return only the plot
Laure Cougnaud
Fisher, R. A. (1936). The Use of Multiple Measurements in Taxonomic Problems. Annals of Eugenics, 7 (2), 179–188
the function used internally: lda
# load data library(ALL) data(ALL) # specify several variables in ldaVar (this might take a few minutes to run...) # sample subsetting: currently cannot deal with missing values samplesToRemove <- which(apply(pData(ALL)[, c("sex", "BT")], 1, anyNA)) # extract random features, because analysis is quite time consuming retainedFeatures <- sample(featureNames(ALL), size = floor(nrow(ALL)/5)) # create the plot esetLda(eset = ALL[retainedFeatures, -samplesToRemove], ldaVar = "BT", colorVar = "BT", shapeVar = "sex", sizeVar = "age", title = "Linear discriminant analysis on the ALL dataset")