sigCheckAll: Run a default set of checks on a gene signature.
In SigCheck: Check a gene signature's prognostic performance against random signatures, known signatures, and permuted data/metadata

Description Usage Arguments Details Value Author(s) References See Also Examples

High-level function for package SigCheck that runs a default set of checks against a predictive signature.

1
2
3

sigCheckAll(check,
            iterations=10, known="cancer", 
            plotResults=TRUE, ...)

`check`	A `SigCheckObject`, as returned by `sigCheck`.
`iterations`	Number of iterations to run to generate background distributions. This is how many random signatures the primary signature will be compared to, as well as how many of each type of permuted dataset will be generated for comparison.
`known`	Specification of a set of known (previously identified) signatures to compare to. See `sigCheckKnown` for more details.
`plotResults`	By default, plots of the results will be generated unless this is set or `FALSE`.
`...`	Extra parameters to pass through `sigCheckPlot`.

This high-level function will run four checks, plot the results, and return a consolidated result set.

First, it calls sigCheckRandom to compare the performance of interations randomly selected signatures.

Next, it calls sigCheckKnown to check the performance of the signature against a database of signatures previously identified to discriminate in other domains.

Finally, two calls are made to sigCheckPermuted to check the performance of randomly permuted metadata and expression data. The first call permutes the survival data if they are available (toPermute="survival"); otherwise it permutes the category assignments (toPermute="categories") The second call permuted the expression value for each gene (permuting each row in the ExpressionSet, equivalent to toPermute="features").

If plotResults is TRUE, the results are plotted. If a classifier is involved, a set of four classification results are plotted in a 2x2 grid, showing how the classification performance of the main signature compares to that of a mode classifier and to the distribution of performance values observed for the random and known signature sets, as well as how it performs using the two type of permuted dataset. If survival data is available, another 2x2 grid is plotted showing how the baseline survival p-value compares to a p-value of 0.05 and to the distribution of p-values observed for the random and known signatures, as well as for the permuted data.

A list containing four elements, each containing the result of a check.

$checkRandom is the result list returned by sigCheckRandom.
$checkKnown is the result list returned by sigCheckKnown.

The third element of the result list will be one of the following:
$checkPermutedSurvival is the result list returned by sigCheckPermuted with toPermute="survival".
$checkPermutedCategories is the result list returned by sigCheckPermuted with toPermute="categories".

The fourth element of the list will be:
$checkPermutedFeatures is the result list returned by sigCheckPermuted with toPermute="features".

Rory Stark

Venet, David, Jacques E. Dumont, and Vincent Detours. "Most random gene expression signatures are significantly associated with breast cancer outcome." PLoS Computational Biology 7.10 (2011): e1002240.

sigCheck, sigCheckRandom, sigCheckPermuted, sigCheckKnown, sigCheckPlot

#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 20, 1000 for real checks

## survival analysis
check <- sigCheck(nki, classes="e.dmfs", survival="t.dmfs",
                  signature=knownSignatures$cancer$VANTVEER,
                  annotation="HUGO.gene.symbol",
                  validationSamples=150:319)
                  
results <- sigCheckAll(check,iterations=ITERATIONS, 
                       known=knownSignatures$cancer[1:20])                  


## classification analysis
check <- sigCheck(nki, classes="e.dmfs", 
                  signature=knownSignatures$cancer$VANTVEER,
                  annotation="HUGO.gene.symbol",
                  validationSamples=275:319,
                  scoreMethod="classifier")
                  
results <- sigCheckAll(check,iterations=ITERATIONS, 
                       known=knownSignatures$cancer[1:20])