RPPAFit-class {SuperCurve}R Documentation

The RPPAFit Class

Description

Objects of the RPPAFit class represent the results of fitting a four-parameter logistic model to the dilution series in a reverse-phase protein array experiment.

Usage

## S4 method for signature 'RPPAFit':
summary(object, ...)
## S4 method for signature 'RPPAFit':
coef(object, ...)
## S4 method for signature 'RPPAFit':
coefficients(object, ...)
## S4 method for signature 'RPPAFit':
fitted(object, type=c("Y", "y", "X", "x"), ...)
## S4 method for signature 'RPPAFit':
residuals(object, type=c("raw", "standardized", "linear"), ...)
## S4 method for signature 'RPPAFit':
resid(object, ...)
## S4 method for signature 'RPPAFit':
image(x, type=c("Residuals", "StdRes", "LinRes","X", "Y"), ...)
## S4 method for signature 'RPPAFit':
hist(x, type=c("Residuals", "StdRes", "LinRes"),
  xlab = NULL, main = NULL, ...)
## S4 method for signature 'RPPAFit':
plot(x, y, type=c("cloud", "series", "individual"),
xlab="Log Concentration", ylab="Intensity", colors=NULL, ...)

Arguments

object A RPPAFit object.
x A RPPAFit object.
type A list of options describing the type of fitted values, residuals, images, histograms, or plots.
xlab Graphics parameter; how the x-axis should be labeled.
ylab Graphics parameter; how the y-axis should be labeled.
main Character string used as a title for the plot.
y not used.
colors A graphical parameter, used only if type='series', to color the lines connecting different dilution series. Eight default colors are used if the argument is NULL.
... The usual extra arguments for generic or plotting routines.

Details

The RPPAFit class holds the results of fitting a four-parameter joint logistic model to all the dilution series on a reverse-phase protein array. For details on how the model is fit, see the RPPAFit function. The basic mathematical model is given by

Y = α + β*g(gamma*(X+delta_i)),

where Y is the observed intensity and X is the designed dilution step. The heart of the model is the function g(x) = e^x/(1+e^x), which is the inverse of the logistic function

f(x) = log(p/(1-p)).

By fitting a joint model, we assume that the parameters α, β, and gamma are the same for all dilution series on the array. The real point of the model, however, is to be able to draw inferences on the delta_i, which represent the (log) concentrations of the protein present in different dilution series.

Value

The summary method has no return value.
The coef and coefficients methods return a named vector of length three.
The fitted method returns a numeric vector.
The resid and residuals methods return a numeric vector.
The image method invisibly returns the object x on which it was invoked.
The hist method returns an object of the histogram class.
The plot method invisibly returns the object x on which it was invoked.

Objects from the Class

Objects should be constructed using the RPPAFit function.

Slots

call:
A call object: the function call that was used to generate this model fit.
rppa:
The RPPA object containing the raw data that was fit
design:
The RPPADesign object describing the layout of the array.
measure:
A character string containing the name of the measurement column in the raw data that was fit by the model.
method:
A character string containing the name of the method that was used to estimate the upper and lower limit parameters in the model.
coefficients:
A named list containing the estimates of model parameters alpha, beta, and gamma.
concentration:
A vector of estimates of the relative log concentration of protein present in each sample.
lower:
A vector containing the lower bounds on the confidence interval of the log concentration estimates.
upper:
A vector containing the upper bounds on the confidence interval of the log concentration estimates.
conf.width:
The width of the confidence interval.
intensities:
The predicted observed intensity at the estimated concentrations for each dilution series.
linear.p:
A vector of p-values for how well a linear model fits each dilution series
ss.ratio:
A statistic measuring the goodness-of-fit for each dilution series, computed as a ratio of two different sums-of-squares.
warn:
A character vector containing any warnings that arose when trying to fit the model to individual dilution series.
version:
A character string containing the version of SuperCurve that produced the fit

Methods

summary(object, ...)
Print a summary of the RPPAFit object. At present, this only reports the function call used to fit the model.
coefficients(object, ...)
Extract a named vector (NOT a list) containing the coefficients alpha, β, and gamma of the model.
coef(object, ...)
An alias for coefficients.
fitted(object, type=c("Y", "y", "X", "x"), ...)
Extract the fitted values of the model. This process is more complictaed than it may seem at first, since we are estimating values on both the X and Y axes. By default, the fitted values are assumed to be the intensities, Y, which are obtained using either an uppercase or lowercase 'y' as the type argument. The fitted log concentrations are returned when type is set to either uppercase or lowercase 'x'. In the notation used above to describe the model, these fitted values are given by X_i = X + delta_i.
residuals(object, type=c("raw", "standardized", "linear"), ...)
Report the residual errors. The 'raw' residuals are defined as the difference between the observed intensities and the fitted intensities, as computed by the fitted function. The 'standardized' residuals are obtained by standardizing the raw residuals. The 'linear' residuals, by contrast, arise from a reinterpretation of the model. Converting from the observed intensity scale by a logistic transformation yields

f((Y - α)/β) = gamma*(X + delta_i) = gamma*X_i.

To compute the linear residuals, we use the observed Y values on the left side and the fitted X values on the right side and take the difference.

resid(object, ...)
An alias for residuals.
image(x, type=c("Residuals", "StdRes", "LinRes","X", "Y"), ...)
The image method produces a 'geographic' plot of the residuals or of the fitted values, depending on the value of the type argument. The implementation reuses code from the image method for an RPPA object.
hist(x, type=c("Residuals", "StdRes", "LinRes"), xlab = NULL, main = NULL, ...)
The hist method produces a histogram of the residuals. The exact form of the residuals being displayed depends on the value of the type argument.
plot(x, y, type=c("cloud", "series", "individual"), xlab="Log Concentration", ylab="Intensity", ...)
The plot method produces a diagnostic plot of the model fit. The default type, 'cloud', simply plots the fitted X values against the observed Y values as a cloud of points around the jointly estimated sigmoid curve. The 'series' plot uses different colored lines to join points belonging to the same dilution series. The 'individual' plot produces separate graphs for each dilution series, laying each one alongside the jointly fitted sigmoid curve.

Author(s)

Kevin R. Coombes <kcoombes@mdanderson.org>

References

See Also

RPPAFit, RPPA, RPPADesign, hist

Examples

path <- system.file("rppaTumorData", package="SuperCurve")
erk2 <- RPPA("ERK2.txt", path=path)
design <- RPPADesign(erk2, grouping="blockSample",
                     controls=list("neg con", "pos con"))
fit.nls <- RPPAFit(erk2, design, "Mean.Net")
image(fit.nls, measure="Residuals")
image(fit.nls, measure="LinRes")
plot(fit.nls, type="cloud")

fit.q <- RPPAFit(erk2, design, "Mean.Net", method="quantiles")
hist(fit.q, type="StdRes")
plot(fit.q, type="series")

coef(fit.nls)
coef(fit.q)

plot(fitted(fit.q), resid(fit.q))
[Package SuperCurve version 0.931 Index]