# R macros for performing mixture 
# modelling of genes and samples with 
# microarray data
# Note: these macros require the mclust and mva libraries;
#       both can be downloaded from the following URL:
#       
#       http://cran.r-project.org/      


ClusterGenes <- function(data,covstruct1="VI",ngroups = 40,covstruct2="VI",K=1000) {
# data should be in the form of a n x m matrix,
# where n is the number of genes and m is the number
# of experimental conditions (i.e. number of genechips)
#
# K = number of groups to cluster into using k-means clustering
#
# covstruct1 is the covariance structure to use in the 
# hierarchical agglomeration step
#
# covstruct2 is the covariance structure to use in the 
# expectation-maximization (EM) algorithm estimation procedure

kmg1 <- kmeans(data,K)
mht1 <- mhtree(data,modelid=covstruct1,partition=kmg1$cluster)
mhc1 <- mhclass(mht1,ngroups)
tmp <- me(data,modelid=covstruct2,ctoz(mhc1[,1]))
list(out=tmp)
}


clustersamps <- function(data,covstruct="VI",nprincomp=2) {
# data should be in the form of a n x m matrix,
# where n is the number of genes and m is the number
# of experimental conditions (i.e. number of genechips)
#

pcx <- prcomp(t(data))$x
bic1 <- emclust(pcx[,1:nprincomp],nclus=1:6,modelid=covstruct)
tmp <- summary(bic1,cmm.pcx[,1:nprincomp])
list(out=tmp)
}