eeMWW: eeMWW

In miccec/yaGST: Competitive gene set and regulon tests.

Description

This function implements the Easy Ensemble, together with the Mann-Witney-Wilcox test, to detect the genes associated with few samples (minority set) being a subset of a larger collection of samples (majority set).

Usage

eeMWW(ddata, minoritySet, runs = 1000)

Arguments

ddata	a matrix where the samples are by rows and the features are in the columns.
minoritySet	a character vector of the minority set matching some row names of ddata.
runs	number of resampling.

Details

The EasyEnsemble (EE) resampling scheme is an Undersampling technique aimed to compare few samples (minority set), carrying some phenotype, to a larger collection of samples (majority set) unrelated with the phenotype. We implement the EE with the Mann-Whitney-Wilcoxon test (MWW) to compare the minority set, of dimension m, with a randomly selected collection of 2*m samples from the majority set.

Value

a named vector of real values.

Note

We suggest running the function in a parallel setup.

Author(s)

Stefano M. Pagnotta

References

Xu-Ying Liu, Jianxin Wu, and Zhi-Hua Zhou - Exploratory Undersampling for Class-Imbalance Learning - IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS?PART B: CYBERNETICS, VOL. 39, NO. 2, APRIL 2009

See Also

mwwGST

Examples

require(yaGST)
nr <- 100; nc <- 1000
# generate a data-matrix with nr samples, and nc features
exprData <- matrix(rpois(nc * nr, 100), nrow = nr, ncol = nc)
colnames(exprData) <- paste0("feat", 1:nc)
rownames(exprData) <- paste0("sam", 1:nr)

# increase the first 3 samples (minority set) of 10% of the original intensity 
# of the first 30 features (later the gene-set)
exprData[1, 1:30] <- exprData[1, 1:30]* runif(30, min = 1, max = 1.10)
exprData[2, 1:30] <- exprData[1, 1:30]* runif(30, min = 1, max = 1.10)
exprData[3, 1:30] <- exprData[1, 1:30]* runif(30, min = 1, max = 1.10)
samples_of_interest <- rownames(exprData)[1:3] # minority set

# running in parallel
library(doParallel)
# adjust the number of CPUs as needed
cl <- makePSOCKcluster(3)
clusterApply(cl, floor(runif(length(cl), max = 10000000)), set.seed)
registerDoParallel(cl)
ans_eeMWW <- eeMWW(exprData, samples_of_interest)
stopCluster(cl)

# set the gene-set and run the enrichment analysis
geneSet <- colnames(exprData)[1:30]
(tmp <- mwwGST(ans_eeMWW, geneSet))
plot(tmp, rankedList = ans_eeMWW)

miccec/yaGST documentation built on May 23, 2019, 7:35 a.m.