order.restricted: Compute the order of a vector subject to restrictions.

Description

When data have been clustered there are generally many different orderings of the rows or columns of the data matrix that are consistent with the clustering information.

This function takes a set of labels, a set of weights and a set of clusters. It returns the labels ordered by increasing weight within clusters. The clusters are ordered by average within cluster weight.

Usage

order.restricted(labels, weights, clusters)

Arguments

labels

The labels, or indices to be ordered.

weights

A vector of weights to be used for the ordering.

clusters

A vector indicating cluster membership.

Value

The input labels ordered by weights within clusters.

Details

Currently all three input vectors are assumed to be the same length. Eventually we would like to extend the functionality of this routine so that it can handle inputs of class dendrogram.

This is a helper function for plotting microarray data. We want to reorder the data so that genes with similar levels of expression (ie those that cluster together) get plotted together.

Other distance measures etc may prove useful.

Examples

Run this code

library(genefilter)
   library(mva)
   data(tdata)
   tdata <- as.matrix(tdata)
   km1 <- kmeans(t(tdata), 3, iter.max=100)
   km2 <- kmeans(tdata, 6, iter.max=100)
   rmns <- apply(tdata, 1, mean)
   cmns <- apply(tdata, 2, mean)
   dn <- dimnames(tdata)
   ord.cols <- order.restricted(dn[[2]], cmns, km1$clust)
   ord.rows <- order.restricted(dn[[1]], rmns, km2$clust)
   tdata <- tdata[ord.rows, ord.cols]