FindSeed: Find highly correlated seed of samples for gene expression...

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

Description

FindSeed() is the key function in MCbiclust. It takes a gene expression matrix and by a stochastic method greedily searches for a seed of samples that maximizes the correlation score of the chosen gene set.

Usage

FindSeed(gem, seed.size, iterations, initial.seed = NULL, messages = 100)

Arguments

gem	Gene expression matrix with genes as rows and samples as columns
seed.size	Size of sample seed
iterations	Number of iterations
initial.seed	Initial seed used, if NULL randomly chosen
messages	frequency of progress messages

Details

Additional options allow for the search to start at a chosen seed, for instance if a improvement to a known seed is desired. The result of FindSeed() is dependent on the number of iterations, with above 1000 usually providing a good seed, and above 10000 an optimum seed.

Value

Highly correlated seed

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)

compmedlab/MCbiclust documentation built on March 9, 2020, 12:14 a.m.