cancerclass (version 1.16.0)

GOLUB: GOLUB DATA

Description

Gene expression data from the leukemia microarray study of Golub et al. [1]. Dataset GOLUB has a dimention of 7129 genes in 72 tumors samples. Dataset GOLUB1 has a dimention of 3571 genes in 72 tumors samples. This dataset is filtered and preprocessed as described in [2].

Usage

data(GOLUB) data(GOLUB1)

Value

Data and annotations are organized in a ExtressenSet of the package Biobase.

GOLUB

ExpressionSet (7129 genes in 72 tumors)

GOLUB1

ExpressionSet (3571 genes in 72 tumors)

References

[1] Golub TR et al (1999), Molecular Classification of cancer: class Discovery and Class Prediction by gene expression monitoring, Science 286:531-7. [2] Dudoit S, Fridlyand J (2002), A prediction-based resampling method for estimating the number of clusters in a dataset, Genome Biol. 3(7):RESEARCH0036.

Examples

### nvalidate
data(GOLUB1)
nval <- nvalidate(GOLUB1[1:200, ])
# Use only the first 200 genes for speed-up of the calculations
plot(nval, type="xy")
plot(nval, type="genes")
plot(nval, type="samples")

### validate
data(GOLUB1)
val <- validate(GOLUB1[1:200, ])
# Use only the first 200 genes for speed-up of the calculations
plot(val, type="xy")
plot(val, type="genes")
plot(val, type="samples")

### fit und predict
data(GOLUB1)
train <- GOLUB1[, 1:38]
test <- GOLUB1[, 39:72]
predictor <- fit(train, method="welch.test")
prediction <- predict(predictor, test, positive="AML", ngenes=50, dist="cor")
plot(prediction, type="histogram", score="zeta")
plot(prediction, type="curves", score="zeta")
plot(prediction, type="roc", score="zeta")
summary(prediction)

### loo
data(GOLUB1)
cv <- loo(GOLUB1, positive="AML", ngenes=10, method="welch.test", dist="cor")
plot(cv, type="histogram", score="zeta")
plot(cv, type="samples", score="zeta")
plot(cv, type="curves", score="zeta")
plot(cv, type="roc", score="zeta")