powercalc {SuperCurve}R Documentation

Power Calculations for a Model Fit to Dilution Series

Description

This function estimates the power to detect a difference in the mean log concentration for two groups of samples in a reverse-phase protein array experiment.

Usage

powercalc(n, dilnFit, upper = 0.95, lower = 0.50, nrep = 1000, alpha = 0.05)

Arguments

dilnFit A RPPAFit object representing the result of fitting a four-parameter logistic model
n size of the sample to be drawn
upper quantile of the concentration on the upper end of the curve
lower quantile fo the concentraion on the lower end of the curve
nrep the number of simulations to perform in order to estiamte the power
alpha significance level at which the power is computed

Details

The powercalc function estimates the power to detect a difference between the upper and lower quantiles of log concentration at the specified significance level α, assuming the samples come from two groups each of size n. The power calculation is based on a Wilcoxon rank sum test. The computation assumes that the only variability arises from the measurement error, as estimated by the residuals from the RPPAFit; thus, it will overestimate the power if there is additional biological variability within the two groups.

Value

Returns a real number, the power.

Author(s)

Kevin R. Coombes <kcoombes@mdanderson.org>

References

KRC

See Also

RPPAFit-class

Examples

path <- system.file("rppaCellData", package="SuperCurve")
akt <- RPPA("Akt.txt", path=path)
design <- RPPADesign(akt, grouping="blockSample",
                     controls=list("neg con", "pos con"))
fit.nls <- RPPAFit(akt, design, "Mean.Net")
# Warning: this takes a while!
powercalc(10, fit.nls, upper=0.75, lower=0.25)

[Package SuperCurve version 0.93 Index]