clipper (version 1.12.0)

clipperAllRoots: Dissect the pathway to find the path with the greatest association with phenotype.

Description

Basing on either variance or mean clique test, this function identifies the paths that are mostly related with the phenotype under study.

Usage

clipperAllRoots(expr, classes, graph, method=c("variance","mean",
"both", "paired"), nperm=100, alphaV=0.05, b=100, trZero=0.001, signThr=0.05,
maxGap=1, permute=TRUE, alwaysShrink=FALSE)

Arguments

expr

an expression matrix or ExpressionSet with colnames for samples and row name for genes.

classes

vector of 1,2 indicating the classes of samples (columns).

graph

a graphNEL object.

method

the kind of test to perform on the cliques. It could be mean, variance, mixed (the best between variance and mean) or paired mean.

nperm

number of permutations. Default = 100.

alphaV

pvalue threshold for variance test to be used during mean test. Default = 0.05.

b

number of permutations for mean analysis. Default = 100.

trZero

lowest pvalue detectable. This threshold avoids that -log(p) goes infinite.

signThr

significance threshold for clique pvalues.

maxGap

allow up to maxGap gaps in the best path computation. Default = 1.

permute

always performs permutations in the concentration matrix test. If FALSE, the test is made using the asymptotic distribution of the log-likelihood ratio. This option should be use only if samples size is >=40 per class.

alwaysShrink

always perform the shrinkage estimates of variance.

Value

roots-paths.Columns are organized as follows:

1. Index of the starting clique
2. Index of the ending clique
3. Index of the clique where the maximum value is reached
4. Length of the path
5. Maximum score of the path
6. Average score along the path
7. Percentage of path activation
8. Impact of the path on the entire pathway
9. Cliques involved and significant
10. Cliques forming the path
11. Genes forming the significant cliques
12. Genes forming the path

Details

The both method combines the results obtained from the mean and variance test. In particular it assign to the cliques the minimum of mean and variance p-values.

References

Martini P, Sales G, Massa MS, Chiogna M, Romualdi C. Along signal paths: an empirical gene set approach exploiting pathway topology. NAR. 2012 Sep.

Massa MS, Chiogna M, Romualdi C. Gene set analysis exploiting the topology of a pathway. BMC System Biol. 2010 Sep 1;4:121.

See Also

cliqueVarianceTest, cliqueMeanTest, getJunctionTreePaths

Examples

if (require(graphite) & require(ALL)){
  kegg  <- pathways("hsapiens", "kegg")
  graph <- pathwayGraph(convertIdentifiers(kegg$'Chronic myeloid leukemia', "entrez"))
  genes <- nodes(graph)
  data(ALL)
  all <- ALL[1:length(genes),1:20]
  classes <- c(rep(1,10), rep(2,10))
  featureNames(all@assayData)<- genes
  graph <- subGraph(genes, graph)
  clipped <- clipperAllRoots(all, classes, graph, "var", trZero=0.01, permute=FALSE)
  clipped[,1:5]
}