supcluster (version 1.0.1)

concordmap: Calculate the Frequency with which each Pair of Features are in the Same Cluster

Description

Label switching is a problem in interpreting the results of a cluster analysis that uses MCMC. Two clusterings may be the same but the labels of the clusters may change. In order to avoid this problem we create a square matrix with length and width equal to the number of features. The i,jth element is the proportion of times feature i and j are in the same cluster. A sorting algorythm puts the genes that are clustered together next to each other.

Usage

concordmap(supcluster.list, chains=1, sort.genes = FALSE,criteria=1)

Arguments

supcluster.list

The output of supcluster

chains

The chains to use in the clustering

sort.genes

If TRUE Genes that associated are put next to each other

criteria

Two genes are in the same cluster when the probability that they are in the same cluster is greater or equal to the criteria.

Value

If sort.genes=TRUE a three element list, the first element is a m x m matrix where m is the number of features and the second element is the ordering created by sorting algorythm that this matrix is in. The final element is the cluster membership for each of the genes. If sort genes=FALSE only the m x m matrix is returned.

See Also

supcluster,,compare.chains,beta.by.gene

Examples

# NOT RUN {
##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.
#NOTE: only a small number of MCMC iterations are done due to time constraints
dat=generate.cluster.data(.2,npats=40,clusts=c(12,8,5),
                sig=1,gamma=1,beta=c(-5,0,6))[[1]]
us=supcluster(dat,outcome="outcome",features=1:25,maxclusters=4,nstart=20,n=40,nchains=2)
ts1=concordmap(us,chains=1)
#plot of the concord map
image(1:25,1:25,ts1$map)

# }