SilhouetteClustGroups: Slihouette validation of correlation vector clusters

In rbentham/MCbiclust: Massive correlating biclusters for gene expression data and associated methods

Slihouette validation of correlation vector clusters

Description

MCbiclust is a stochastic method and needs to be run multiple times to identify different biclusters. SilhouetteClustGroups() examines the correlation vectors calculated from different runs and uses the technique of examining silhouette widths to identify the number of distinct clusters (and hence biclusters) found.

Usage

SilhouetteClustGroups(cor.vec.mat, max.clusters, plots = FALSE, seed1 = 100,
  rand.vec = TRUE)

Arguments

cor.vec.mat	Correlation matrix of the correlation vectors (CVs)
max.clusters	Maximum number of clusters to divide CVs into
plots	True or False for whether to show silhouette plots
seed1	Value used to set random seed
rand.vec	True or False for whether to add random correlation vector used for comparison

Value

The distinct clusters of correlation vectors

Examples

data(CCLE_small)
data(Mitochondrial_genes)

mito.loc <- (row.names(CCLE_small) %in% Mitochondrial_genes)
CCLE.mito <- CCLE_small[mito.loc,]

CCLE.seed <- list()
CCLE.cor.vec <- list()

for(i in 1:5){
    set.seed(i)
    CCLE.seed[[i]] <- FindSeed(gem = CCLE.mito,
                               seed.size = 10,
                               iterations = 100,
                               messages = 100)}

for(i in 1:5){
  CCLE.cor.vec[[i]] <-  CVEval(gem.part = CCLE.mito,
                               gem.all = CCLE_small,
                               seed = CCLE.seed[[i]],
                               splits = 10)}

CCLE.cor.mat <- as.matrix(as.data.frame(CCLE.cor.vec))

CCLE.clust.groups <- SilhouetteClustGroups(cor.vec.mat = CCLE.cor.mat,
                                           plots = TRUE,
                                           max.clusters = 10)

av.corvec.fun <- function(x) rowMeans(CCLE.cor.mat[,x])
CCLE.average.corvec <- lapply(X = CCLE.clust.groups,
                              FUN = av.corvec.fun)

multi.sort.prep <- MultiSampleSortPrep(gem = CCLE_small,
                                       av.corvec = CCLE.average.corvec,
                                       top.genes.num = 750,
                                       groups =CCLE.clust.groups,
                                       initial.seeds = CCLE.seed)  

multi.sort <- list()        
for(i in seq_len(length(CCLE.clust.groups))){
    multi.sort[[i]] <- SampleSort(multi.sort.prep[[1]][[i]],
                                  seed = multi.sort.prep[[2]][[i]],
                                  sort.length = 11)
}

rbentham/MCbiclust documentation built on Feb. 5, 2024, 7:44 a.m.