claudinLow: Claudin-low classification for Breast Cancer Data

In bhklab/genefu: Computation of Gene Expression-Based Signatures in Breast Cancer

Claudin-low classification for Breast Cancer Data

Description

Subtyping method for identifying Claudin-Low Breast Cancer Samples. Code generously provided by Aleix Prat.

Usage

claudinLow(x, classes="", y, nGenes="", priors="equal",
  std=FALSE, distm="euclidean", centroids=FALSE)

Arguments

x	the data matrix of training samples, or pre-calculated centroids.
classes	a list labels for use in coloring the points.
y	the data matrix of test samples.
nGenes	the number of genes selected when training the model.
priors	'equal' assumes equal class priors, 'class' calculates them based on proportion in the data.
std	when true, the training and testing samples are standardized to mean=0 and var=1.
distm	the distance metric for determining the nearest centroid, can be one of euclidean, pearson, or spearman.
centroids	when true, it is assumed that x consists of pre-calculated centroids.

Value

A list with items:

• predictions

• testData

• distances

• centroids

References

Aleix Prat, Joel S Parker, Olga Karginova, Cheng Fan, Chad Livasy, Jason I Herschkowitz, Xiaping He, and Charles M. Perou (2010) "Phenotypic and molecular characterization of the claudin-low intrinsic subtype of breast cancer", Breast Cancer Research, 12(5):R68

See Also

medianCtr(), q

Examples

data(claudinLowData)

#Training Set
train <- claudinLowData
train$xd <-  medianCtr(train$xd)
# Testing Set
test <- claudinLowData
test$xd <-  medianCtr(test$xd)

# Generate Predictions
predout <- claudinLow(x=train$xd, classes=as.matrix(train$classes$Group,ncol=1), y=test$xd)

# Obtain results
results <- cbind(predout$predictions, predout$distances)
#write.table(results,"T.E.9CELL.LINE_results.txt",sep="\t",col=T, row=FALSE)

bhklab/genefu documentation built on Nov. 30, 2024, 9:03 p.m.