sigCheckPlot: Plot results of a signature check or set of checks

In SigCheck: Check a gene signature's prognostic performance against random signatures, known signatures, and permuted data/metadata

Description

Plots results of a signature check, as returned by sigCheckRandom, sigCheckKnown, sigCheckPermuted, or sigCheckAll.

Usage

sigCheckPlot(checkResults, classifier=FALSE, 
             title, nolegend=FALSE, ...)

Arguments

checkResults	The list returned by sigCheckRandom, sigCheckKnown, sigCheckPermuted, or sigCheck.
classifier	If a classifier was used in the original call to sigCheck, setting this to TRUE will result in a plot showing how the primary signature compares based on classification performance (rather than survival).
title	Title string for plot. If missing, a default plot title will be generated.
nolegend	IF TRUE, no legend will be included with the plot(s).
...	Additional arguments to be passed to the plot function.

Details

For results based on survival analysis, the background distribution of p-values (in 1-log10() format) derived from the check (either random signatures, known signatures, or performance using permuted data) is plotted. Up to two vertical red lines are also plotted: a solid red line representing the performance of the primary signature/data, and a dotted red line representing a p-value of 0.05. One or both of these may be missing if their performance falls outside the range of the background distributions.

For results based on classification performance, the x-axis represents the range of classification performance scores computed in the check, and the y-axis representing how many times that score was obtained. In addition, vertical lines are plotted representing the classification performance of the originally specified signature (solid red line) and the performance of a classifier that always predicts the mode value of the training samples (dotted red line).

If the results of sigCheckAll is passed in, all four results plots are generated in a 2x2 grid.

Value

none

Note

Better formance of the signature being checked results in the solid red line being to the right of the background distribution. For survival results, this indicates a lower-value. For classification results,this indicates superior classification performance.

Author(s)

Rory Stark with Justin Norden

See Also

sigCheck, sigCheckAll, sigCheckRandom, sigCheckKnown, sigCheckPermuted

Examples

#Disable parallel so Bioconductor build won't hang
library(BiocParallel)
register(SerialParam())

library(breastCancerNKI)
data(nki)
nki <- nki[,!is.na(nki$e.dmfs)]
data(knownSignatures)
ITERATIONS <- 5 # should be at least 1000 for real checks

## survival analysis with separate training and validation using SVM
check <- sigCheck(nki, classes="e.dmfs", survival="t.dmfs",
                  signature=knownSignatures$cancer$VANTVEER,
                  annotation="HUGO.gene.symbol",
                  validationSamples=250:319,
                  scoreMethod="classifier", threshold=.33) 
                  
results <- sigCheckRandom(check,iterations=ITERATIONS)
par(mfrow=c(1,2))
sigCheckPlot(results)
sigCheckPlot(results, classifier=TRUE)

SigCheck documentation built on Nov. 8, 2020, 6:38 p.m.