IntClust (version 0.1.0)

TrackCluster: Follow a cluster over multiple methods

Description

It is often desired to track a specific selection of object over the different methods and/or weights. This can be done with the ClusterDistribution. For every method, it is tracked where the objects of the selections are situated.

Usage

TrackCluster(List, Selection, nrclusters = NULL, followMaxComps = FALSE,
  followClust = TRUE, fusionsLog = TRUE, weightclust = TRUE,
  names = NULL, selectionPlot = FALSE, table = FALSE,
  completeSelectionPlot = FALSE, ClusterPlot = FALSE, cols = NULL,
  legendposx = 0.5, legendposy = 2.4, plottype = "sweave",
  location = NULL)

Arguments

List

A list of the clustering outputs.The first element of the list will be used as the reference in ReorderToReference.

Selection

The selection of objects to follow or a specific cluster number. Default is NULL.

nrclusters

The number of clusters to cut the dendrogram in. Default is NULL.

followMaxComps

Logical for plot. Whether to follow the maximum of objects. Default is FALSE.

followClust

Logical for plot. Whether to follow the specific cluster. Default is TRUE.

fusionsLog

Logical. To be handed to ReorderToReference: indicator for the fusion of clusters. Default is TRUE

weightclust

Logical. To be handed to ReorderToReference: to be used for the outputs of CEC, WeightedClust or WeightedSimClust. If TRUE, only the result of the Clust element is considered. Default is TRUE.

names

Optional. Names of the methods. Default is NULL.

selectionPlot

Logical. Should a plot be produced. Depending on followMaxComps and followClust it focuses on the maximum of objects or a cluster. It will not be indicated to which cluster objects moved. Default is FALSE.

table

Logical. Should a table with the objects per method and the shared objects be produced? Default is FALSE.

completeSelectionPlot

Logical. Should the complete distribution of the selection be plotted? This implies that it will be indicated to which cluster objects will move. Default is FALSE.

ClusterPlot

Logical. Plot of specific cluster. Default is FALSE.

cols

The colors used for the different clusters. Default is NULL.

legendposx

The x-coordinate of the legend on all plots. Default is 0.5.

legendposy

The y-coordinate of the legend on all plots. Default is 2.4.

plottype

Should be one of "pdf","new" or "sweave". If "pdf", a location should be provided in "location" and the figure is saved there. If "new" a new graphic device is opened and if "sweave", the figure is made compatible to appear in a sweave or knitr document. Default is "new".

location

If plottype is "pdf", a location should be provided in "location" and the figure is saved there. Default is NULL.

Value

The returned value is a list with an element for every method. This element is another list with the following elements:

Selection

The selection of objects to follow

nr.clusters

the number of clusters the selection is divided over

nr.min.max.together

the minimum and maximum number of objects found together

perc.min.max.together

minimum and maximum percentage of objects found together

AllClusters

A list with an element per cluster that contains at least one of the objects in Selection. The list contains the cluster number, the complete cluster, the objects that originally could be found in this cluster and which object were joined extra to it.

Depending on whether followMaxComps or followClust is specified, the cluster of interest is mentioned separately as well for easy access. If the option was specified to create a table, this can be found under the Table element. Each plot that was specified to be created is plotted in a new window in the graphics console.

Details

The result is provided with extra information as which objects of the original selection can be found in this cluster and which are extra. Further, plots of the distribution of the objects can be produced. One plot follows the complete distribution of the cluster while another one focuses on either the maximum number of objects or a specific cluster, whatever is specified. It are the number of objects that are plotted and the first element indicated the number of objects in the selection. A table can be produced as well, that separates the objects that are shared over all methods from those extra in the original selection and extra for the other methods. The ReorderToReference is applied to make sure that the clusters are comparable over the methods.

The function is experimental and might not work in specific cases. Please let us know such that we can improve its functionality.

Examples

# NOT RUN {
data(fingerprintMat)
data(targetMat)
data(Colors1)

MCF7_F = Cluster(fingerprintMat,type="data",distmeasure="tanimoto",normalize=FALSE,
method=NULL,clust="agnes",linkage="flexible",gap=FALSE,maxK=55,StopRange=FALSE)
MCF7_T = Cluster(targetMat,type="data",distmeasure="tanimoto",normalize=FALSE,
method=NULL,clust="agnes",linkage="flexible",gap=FALSE,maxK=55,StopRange=FALSE)

L=list(MCF7_F,MCF7_T)
names=c("FP","TP")

Comps=FindCluster(List=L,nrclusters=7,select=c(1,4))
Comps

CompsFPAll=TrackCluster(List=L,Selection=Comps,nrclusters=7,followMaxComps=TRUE,
followClust=FALSE,fusionsLog=TRUE,weightclust=TRUE,names=names,selectionPlot=TRUE,
table=TRUE,completeSelectionPlot=TRUE,cols=Colors1,plottype="new",location=NULL)
# }
# NOT RUN {
# }