vsn (version 3.40.0)

vsnh: A function that transforms a matrix of microarray intensities.

Description

A function that transforms a matrix of microarray intensities. This function works in conjunction with vsn. vsn and vsnh have been superseded by vsn2 and the predict method for vsn objects. The functions vsn and vsnh remain in the package for backward compatibility, but for new projects, please use vsn2 and predict.

Usage

vsnh(y, p, strata)

Arguments

y

A numeric matrix containing intensity values from an array experiment. It may contain NA values.

p

An array with the transformation parameters. If strata is specified, it must be a 3d array, dim(p)[1] must be greater than or equal to the maximum of strata, dim(p)[2] must be ncol(y), and dim(p)[3] must be 2. If strata is missing, then the first dimension may be omitted. NA values are not allowed. See Details.

strata

Integer vector of length nrow(y). See vsn for details.

Value

A numeric matrix of the same size as y, with the transformed data.

Details

The transformation is: vsnh(y, p, s)[k, i] = asinh( p[s[k], i, 1] + p[s[k], i, 2] * y[k, i] ) - log(2*p[s[1], 1, 2]) where k=1:nrow(y) counts over the probes, i=1:ncol(y) counts over the samples, p[s[k], i, 1] is the calibration offset for stratum s[k] in sample i, p[s[k], i, 2] is the calibration factor for stratum s[k] in sample i, and s[k] is the stratum of the the k-th probe.

The constant offset - log(2*p[s[1], 1, 2]) is there to make sure that for large y, vsnh(y) for the first stratum on the first chip is approximately the same as log(y). This has no effect on the generalized log-ratios (glog-ratios), which are differences between transformed intensities, but some users are more comfortable with the absolute values that are obtained this way, since they are more comparable to the log scale.

References

Variance stabilization applied to microarray data calibration and to the quantification of differential expression, Wolfgang Huber, Anja von Heydebreck, Holger Sueltmann, Annemarie Poustka, Martin Vingron; Bioinformatics (2002) 18 Suppl.1 S96-S104.

Parameter estimation for the calibration and variance stabilization of microarray data, Wolfgang Huber, Anja von Heydebreck, Holger Sueltmann, Annemarie Poustka, and Martin Vingron; Statistical Applications in Genetics and Molecular Biology (2003) Vol. 2 No. 1, Article 3. http://www.bepress.com/sagmb/vol2/iss1/art3.

See Also

vsn

Examples

data(kidney)
y      = exprs(kidney)
p      = array(c(-0.2, -0.1, 0.1, 0.2, 0.0026, 0.0028, 0.0030, 0.0032), dim=c(2,2,2))
strata = sample(1:2, nrow(y), replace=TRUE)
res1   = vsnh(exprs(kidney), p, strata)

res2   = asinh(p[strata,,1] + p[strata,,2] * y) - log(2*p[strata,1,2])

stopifnot(max(abs(res1 - res2)) < 1e-10)