ReorderCluster-package: optimal reordering of the hierarchical tree according to...

In ReorderCluster: Reordering the Dendrogram According to the Class Labels

optimal reordering of the hierarchical tree according to class labels

Description

The package includes the functions to make the near optimal reordering of the leaves of dendrogram according to the class labels, preserving its initial structure. The reordering algorithm tries to group instances from the same class together. With such an optimised dendrogram it is easier to interpret the interrelation between the available class labels and clusters. Moreover, it is possible to see whether one can predict the class for unlabeled instances based on the distance that was used for clustering.

Details

Package:	ReorderCluster
Type:	Package
Version:	1.0
Date:	2014-07-21
License:	GPL (>= 3)

The basic unit of ReorderCluster package is the function RearrangeJoseph, which makes the initialization of all the auxiliary matrices and the sequence of the necessary functions' call to perform the leaf reordering according to class labels. The function RearrangeJoseph implements the preprocessing of the merging matrix of the hierarchical clustering by calling the auxiliary function testBar to form the node (subtree) structure using the merging matrix, the auxiliary functions CalcMerge to mark the nodes with identical class labels, the auxiliary functions SubTree to simplify the initial hierarchical tree by constructing the modified tree, where the subtrees with equal class labels are merged into the single element. The function RearrangeJoseph calls the function OrderingJoseph to execute the dynamic programming algorithm, which calculates the values of the evaluation function for each subtree of the dendrogram and thereafter finds the optimized sequence of leaves using the function funMerge. The evaluation function is calculated on the basis of the existing class labels.

Functions

RearrangeJoseph	Makes the calls to sequence of functions to perform the leaf reordering according to class labels.
OrderingJoseph	Calculates the evaluation function for each subtree using the dynamic programming approach.
OrderingJosephC	Calculates the evaluation function for each subtree using the dynamic programming approach calling the function in C++ language.
RearrangeData	Provides the main scheme to perform optimal reordering of the hierarchical tree according to class labels for any dataset.
funMerge	Recovers the optimal sequence of leaves of the hierarchical tree using the backtracking scheme.
CalcMerge	Forms the auxiliary vectors to mark the nodes with identical class labels.
SubTree	Simplifies the initial hierarchical tree by reducing the number of nodes.
testBar	For each inner node of the dendrogram forms two vectors with left and write subtrees' elements.
colorDendClass	Makes the plot of the dendrogram, visualizing the class label information with different colors of dendrogram edges.
testData1	Generates the simulated dataset with 400 genes and 100 experiments with 3 class labels.
testData2	Generates the simulated dataset with 90 genes and 90 experiments with 3 class labels.

Installing and using

To install this package, make sure you are connected to the internet and issue the following command in the R prompt:

    install.packages("ReorderCluster")
  

To load the package in R:

    library(ReorderCluster)
  

Author(s)

Natalia Novoselova,Frank Klawonn,Junxi wang

Maintainer: Natalia Novoselova <[e-mail]>

References

Z. Bar-Joseph, D.K. Gifford, and T.S. Jaakkola. Fast optimal leaf ordering for hierarchical clustering. Bioinformatics, 17:22-29, 2001. Therese Biedl,Brona Brejova, Erik D,Demaine, Angele M.Hanmel and Tomas Vinar:Optimal Arrangement of Leaves in the Tree Representing Hierarchical Clustering of Gene Expression Data/Technical report 2001-14.

See Also

CRAN packages cba, seriation includes the function for optimal leaf reordering of the dendrogram with respect to the minimizing the sum of the distances along the (Hamiltonian) path connecting the leaves in the given order.

Examples

  data(leukemia)
  cpp=FALSE
  rownames(leukemia)=leukemia[,1]
  leukemia=leukemia[,-1]
  matr=leukemia[,-101]
  class=leukemia[,101]
  
  matr=as.matrix(matr)
  dd=dim(matr)

	label=unique(class)
	
	Rowcolor=rainbow(length(label))
	rc=matrix(0,length(class),1)
	ds=matrix(0,length(class),1)

	for (j in 1:length(label))
  	{
		index=which(class==label[j])
		rc[index]=Rowcolor[j]
	}
	
	dist=dist(matr)
	hc <- hclust(dist)
	
	dend=as.dendrogram(hc)
  
	my_palette <- colorRampPalette(c("green", "black", "red"))(n = 399)
	hv <- heatmap.2(matr,Rowv=dend,scale = "none",Colv=NA,
	                col=my_palette, RowSideColors = rc,trace="none",dendrogram="row")
  legend("topright",legend=label,col=Rowcolor,pch=15,cex=0.8)

  res=RearrangeJoseph(hc,as.matrix(dist),class,cpp) 
  
	hcl=res$hcl

  dend=as.dendrogram(hcl)
  
	hv <- heatmap.2(matr,Rowv=dend,scale = "none",Colv=NA,
	                col=my_palette, RowSideColors = rc,trace="none",dendrogram="row")
  legend("topright",legend=label,col=Rowcolor,pch=15,cex=0.8)

ReorderCluster documentation built on June 21, 2022, 5:05 p.m.