HclustGenesHiCor: Find the most highly correlated genes using hierarchical...

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

Description

Upon finding an initial bicluster with FindSeed() not all the genes in the chosen geneset will be highly correlated to the bicluster. HclustGenesHiCor() uses the output of FindSeed() and hierarchical clustering to only select the genes that are most highly correlated to the bicluster. This is achieved by cutting the dendogram produced from the clustering into a set number of groups and then only selecting the groups that are most highly correlated to the bicluster

Usage

HclustGenesHiCor(gem, seed, cuts)

Arguments

gem	Gene expression matrix with genes as rows and samples as columns
seed	Seed of highly correlating samples
cuts	Number of groups to cut dendogram into

Value

Numeric vector of most highly correlated genes

Examples

data(CCLE_small)
data(Mitochondrial_genes)

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

random.seed <- sample(seq(length = dim(CCLE.mito)[2]),10)
CCLE.seed <- FindSeed(gem = CCLE.mito,
                      seed.size = 10,
                      iterations = 100,
                      messages = 100)


CorScoreCalc(CCLE.mito, random.seed)
CorScoreCalc(CCLE.mito, CCLE.seed)

CCLE.hicor.genes <- as.numeric(HclustGenesHiCor(CCLE.mito,
                                                CCLE.seed,
                                                cuts = 8))

CorScoreCalc(CCLE.mito[CCLE.hicor.genes,], CCLE.seed)

Example output

Iteration	Correlation Score
100 		 0.36178
[1] 0.2871502
[1] 0.3617806
[1] 0.3862503

MCbiclust documentation built on Nov. 8, 2020, 11:09 p.m.