geneSignatureFinder (version 2014.02.17)

goodAndPoorClassification: Function to classify the unsupervised clusters found through a classification procedure in good and poor groups of samples.

Description

This function label the groups found by the unsupervised classification in good and poor.

Usage

goodAndPoorClassification(clustering)

Arguments

clustering

factor of so many elements as the number of samples containing the classification in two groups. Missing values are allowed.

Value

Details

The labelling of the two groups in good and poor is obtained through the comparion of the expected and the observed failures computed in the test statistics of the log-rank test. The group for which the expected are more than the observed are classified are good prognosis. The variable stData has to be defined in the environment.

Examples

data(geNSCLC)
geData <- geNSCLC

data(stNSCLC)
stData <- stNSCLC
######################
randomClustering  <- runif(nrow(stData)) > .5
survdiff(stData ~ randomClustering)
###############
clusteringByPrognosis <- goodAndPoorClassification(randomClustering)
survdiff(stData ~ clusteringByPrognosis)