plotPosAnnotInComp: Histograms of sample contributions for each annotation level

Description

This function plots the positions of groups of samples formed by the variables (i.e the sample annotations) across all the components of an object of class icaSet. For each variable level (e.g for each tumor stage) this function plots the positions of the corresponding samples (e.g the subset of samples having this tumor stage) within the histogram of the global sample contributions. The plots are saved in pdf file, one file is created per variable. The pdf files are names 'variable.pdf' and save either in pathPlot if specified or the current directory.

Usage

plotPosAnnotInComp(icaSet, params, keepVar = varLabels(icaSet), keepComp = indComp(icaSet), keepSamples = sampleNames(icaSet), pathPlot = NULL, breaks = 20, colAll = "grey74", colSel, resClus, funClus = c("Mclust", "kmeans"), nbClus = 2, by = c("annot", "component"), typeImage = c("pdf", "png", "none"), ...)

Arguments

icaSet

An object of class IcaSet

params

A MineICAParams object

keepVar

The variable labels to be considered, i.e a subset of the column labels of the pheno data of icaSet available in (varLabels(icaSet))

keepComp

A subset of components available in indComp(icaSet); by default, all components are used

keepSamples

A subset of samples, must be available in sampleNames(icaSet); by default, all samples are used

pathPlot

A character specifying the path where the plots will be saved

breaks

The number of breaks to be used in the histograms

colSel

The colour of the histogram of the group of interest, default is "red"

colAll

The colour of the global histogram, default is "grey74"

resClus

A list containing the outputs of function clusterSamplesByComp, which consists of sample clustering applied to matrix A of argument icaSet. If missing, the clustering is performed by the function.

funClus

The clustering method to be used, either "Mclust" or "kmeans". If resClus is not missing, equals resClus$funClus.

nbClus

If resClus is missing, it provides the number of clusters to be computed by funClus, default is 2

Either "annot" to plot the histograms of each variable across all components, or "component" to plot the histograms for each component across variables. When by="annot" one pdf file is created by variable name, while when annot="component", one pdf file is created by component.

typeImage

The type of image to be created, either "pdf" (default) or "png". "png" is not recommended, unless there are at the most 4 histograms to be plotted, because it does not allow to deal with multiple pages of plots.

...

Additional parameters for function hist

Value

NULL

Details

The plotted values are the sample contributions across the components, i.e across the columns of A(icaSet).

If argument resClus is missing, the function computes the clustering of the samples on each component (i.e on each column of A(icaSet)) using funClus and nbClus.

The association between the clusters and the considered sample group is tested using a chi-square test. The p-values of these tests are available in the title of each plot.

When by="annot" this function plots the histograms of each variable across all components, to plot the histograms for each component across variables, please use by="component".

Examples

Run this code

## Not run: 
# ## load an example of IcaSet
# data(icaSetCarbayo)
# 
# ## Use icaSetCarbayo, look at the available annotations
# varLabels(icaSetCarbayo)
# 
# ## Plot positions of samples in components according to annotations 'SEX' and 'STAGE'
# # plots are saved in files SEX.pdf and STAGE.pdf created in the current directory
# plotPosAnnotInComp(icaSet=icaSetCarbayo, keepVar=c("SEX","STAGE"), keepComp=1:2,  funClus="Mclust")
# # specifiy arg 'pathPlot' to save the pdf in another directory, but make sure it exists before
# # specifiy arg 'by="comp"' to create one pdf file per component
# ## End(Not run)