pathwaySurvivalTest: Survival analysis on the whole pathway
In cavei/survClip: Survival Analysis for Pathways

Description Usage Arguments Details Value See Also Examples

This function performs survival analysis on pathways. The analysis can be either topological or classical. The analysis is based on data reduction based on Principal Component Analysis and a Cox proportional hazard model on the most influent PCs

pathwaySurvivalTest(expr, survAnnot, graph,
  pcsSurvCoxMethod = c("regular", "topological", "sparse"),
  alwaysShrink = FALSE, maxPCs = 10,
  survFormula = "Surv(days, status) ~", robust = FALSE)

`expr`	expression matrix
`survAnnot`	a data frame for survival annotations specified according to the survFormula. The data frame must contain days and status
`graph`	a graphNEL object for a graph
`pcsSurvCoxMethod`	a method to perform PCA. Can be "regular", "topological", "sparse" for regular PCA, topological based PCA and sparse PCA, respectively. The latter one (sparse) is particularly suited for cliques only
`alwaysShrink`	if TRUE, always shrink the covariance matrix. Deafult=FALSE
`maxPCs`	maximum number of PCs used in the cox formula "Surv(days, status) ~ PC1.."
`survFormula`	the formula used in Coxph analysis. Defaut="Surv(days, status) ~". Please note that the formula end with '~' meaning that PCs will be added
`robust`	should be used the robust mode for cox

With "regular" method, a regular PCA analysis is used to compute PCs. With "topological" method, the covariance matrix is estimated using the topology of the pathway with IPS algorithm. With "sparse" method, a penalized regression is used for the estimation of PCs (as implemented in elasticnet). The max number of PCs used by the model is estimated by "estim_ncp" in FactoMineR. A maximum number of PCs can be fixed by the user. The minimum ot the two is chosen.

A survPath object

cliqueSurvivalTest

if (require(graphite)) {
  data(exp)
  data(survAnnot)
  data(graph)
  row.names(exp) <- paste0("ENTREZID:", row.names(exp))
  genes <- intersect(graph::nodes(graph), row.names(exp))
  graph <- graph::subGraph(genes, graph)
  expr <- exp[genes, , drop=FALSE]
  pathwaySurvivalTest(expr, survAnnot, graph, maxPCs=2)
}