ClassComparison (version 3.1.6)

TNoM-class: Classes "TNoM" and "fullTNoM"

Description

Implements the "Total Number of Misclassifications" method for finding differentially expressed genes.

Usage

TNoM(data, classes, verbose=TRUE)
# S4 method for TNoM
summary(object, …)
# S4 method for TNoM
update(object, nPerm, verbose=FALSE, …)
# S4 method for TNoM
selectSignificant(object, cutoff, …)
# S4 method for TNoM
countSignificant(object, cutoff, …)
# S4 method for fullTNoM,missing
plot(x, y, …)
# S4 method for fullTNoM
hist(x, …)

Arguments

data

Either a data frame or matrix with numeric values or an ExpressionSet as defined in the BioConductor tools for analyzing microarray data.

classes

If data is a data frame or matrix, then classes must be either a logical vector or a factor. If data is an ExpressionSet, then classes can be a character string that names one of the factor columns in the associated phenoData subobject.

verbose

logical scalar. If TRUE, print out intermediate results

object

object of class TNoM

nPerm

integer scalar specifying the number of permutations to perform

cutoff

integer scalar

x

object of class fullTNoM

y

Nothing, since it is supposed to be missing. Changes to the Rd processor require documenting the missing entry.

…

extra arguments to generic or plotting routines

Value

summary returns a TNoMSummary object.

update returns a fullTNoM object.

selectSignificant returns a vector of logical values.

countSignificant returns an integer.

Creating Objects

Although objects of the class can be created by a direct call to new, the preferred method is to use the TNoM generator. The inputs to this function are the same as those used for row-by-row statistical tests throughout the ClassComparison package; a detailed description can be found in the MultiTtest class.

Slots

Objects of the TNoM class have the following slots:

data:

The data matrix used to construct the object

tnomData:

numeric vector, whose length is the number of rows in data, recording the minimum number of misclassification achieved using this data row.

nCol:

The number of columns in data

nRow:

The number of rows in data

classifier:

The classification vector used to create the object.

call:

The function call that created the object

Objects of the fullTNoM class have the following slots:

dex:

Numeric vector of the different possible numbers of misclassifications

fakir:

Numeric vector of expected values based on simulations

obs:

Numeric vector of observed values

scr:

Numeric vector of values based on a permutation test

name:

A character string with a name for the object

Methods

Objects of the TNoM class have the following methods:

summary(object, …)

Write out a summary of the object, including the number of genes achieving each possible number of misclassifications.

countSignificant(object, cutoff, …)

Count the number of significant genes at the given cutoff.

selectSignificant(object, cutoff, …)

Get a vector for selecting the number of significant genes at the given cutoff.

update(object, nPerm, verbose=FALSE, …)

Perform simulation and permutation tests on the TNoM object.

Objects of the fullTNoM class have the following methods:

plot(x, …)

Plot a summary of the TNoM object. This consists of three curves: the observed cumulative number of genes at each misclassification level, along with the corresponding numbers expected based on simulations or permutation tests. The colors of the curves are controlled by the values of oompaColor$OBSERVED, oompaColor$EXPECTED, and oompaColor$PERMTEST

hist(x, …)

Plot a not terribly useful nor informative histogram of the results.

Details

The TNoM method was developed by Yakhini and Ben-Dor and first applied in the melanoma microarray study by Bittner and colleagues (see references). The goal of the method is to detect genes that are differentially expressed between two groups of samples. The idea is that each gene serves as a potential classifier to distinguish the two groups. One starts by determining an optimal cutoff on the expression of each gene and counting the number of misclassifications that gene makes. Next, we bin genes based on the total number of misclassifications. This distribution can be compared with the expected value (by simulating normal data sets of the same size). Alternatively, one can estimate the null distribution directly by scrambling the sample labels to perform a permutation test.

The TNoM constructor computes the optimal cutoffs and the misclassification rates. The update method performs the simulations and permutation tests, producing an object of the fullTNoM class.

References

Bittner M, Meltzer P, Chen Y, Jiang Y, Seftor E, Hendrix M, Radmacher M, Simon R, Yakhini Z, Ben-Dor A, Sampas N, Dougherty E, Wang E, Marincola F, Gooden C, Lueders J, Glatfelter A, Pollock P, Carpten J, Gillanders E, Leja D, Dietrich K, Beaudry C, Berens M, Alberts D, Sondak V. Molecular classification of cutaneous malignant melanoma by gene expression profiling. Nature. 2000 Aug 3;406(6795):536-40.

See Also

Bum, MultiTtest, MultiWilcoxonTest

Examples

# NOT RUN {
showClass("TNoM")
showClass("fullTNoM")
n.genes <- 200
n.samples <- 10

bogus <- matrix(rnorm(n.samples*n.genes, 0, 3), ncol=n.samples)
splitter <- rep(FALSE, n.samples)
splitter[sample(1:n.samples, trunc(n.samples/2))] <- TRUE

tn <- TNoM(bogus, splitter)
summary(tn)

tnf <- update(tn)
plot(tnf)
hist(tnf)
# }