Sigmoidal transformation of the score values stored in a cellHTS object obtaining the call values for each probe.

Share:

Description

Apply a sigmoidal transformation with parameters z0 and lambda to the summarized scored values stored in a cellHTS object. The obtained results are called calls and are stored in slot assayData, overridding its current content.

Currently this function is implemented only for single-color data.

Usage

1
scores2calls(x, z0, lambda)

Arguments

x

an object of class cellHTS containing replicate data that have already been scored and summarized (see details).

z0

a numeric value giving the centre of the sigmoidal transformation. See details.

lambda

a numeric value (>0) that corresponds to the parameter lambda of the sigmoidal transformation. This value should be >0, but usually it makes more sense to use a value >=1. See details.

Details

This function applies a sigmoidal transformation with parameters z0 and lambda to the single per-probe score values stored in a cellHTS object. The obtained results are called calls. The transformation is given by:

1 / (1 + exp(-lambda * (z- z0)))

where z are the score values, z0 is the centre of the sigmoidal transformation, and the lambda is a parameter that controls the smoothness of the transformation. The higher is lambda, more steeper is the transition from lower to higher values. lambda should be > 0, but usually it makes more sense to use a value >=1.

This transformation maps the score values to the interval [0,1], and is intended to expand the scale of scores with intermediate values and shrink the ones showing extreme values, therefore making the difference between intermediate phenotypes larger.

Value

The cellHTS object with the call values stored in slot assayData. This is an object of class assayData corresponding to a single matrix of dimensions Features x 1.

Author(s)

W. Huber huber@ebi.ac.uk, Ligia Braz ligia@ebi.ac.uk

References

Boutros, M., Bras, L.P. and Huber, W. (2006) Analysis of cell-based RNAi screens, Genome Biology 7, R66.

See Also

normalizePlates, summarizeChannels, scoreReplicates, summarizeReplicates, imageScreen.

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
    data(KcViabSmall)
    x <- normalizePlates(KcViabSmall, scale="multiplicative", method="median", varianceAdjust="none")
    x <- scoreReplicates(x, sign="-", method="zscore")
    x <- summarizeReplicates(x, summary="min")
    xc <- scores2calls(x, z0=1.5, lambda=2) 
    plot(Data(x), Data(xc), col="blue", xlab="z-scores", ylab="calls", main=expression(1/(1+e^{-lambda *(z-z[0])})))
    if(require(splots)) {
     sp = split(Data(xc), plate(xc))
     plotScreen(sp, zrange=c(0,1), fill=c("white", "red"), na.fill="yellow",
               main="Calls", ncol=3L)
    }

Want to suggest features or report bugs for rdrr.io? Use the GitHub issue tracker.