selectStatsBonferroni {goCluster} | R Documentation |
This function corrects the p-values for the annotation terms in the given tree by taking the total number of elements in the tree into account. It removes all annotation terms with a p-value higher than the given threshold after the values have been corrected.
selectStatsBonferroni(X, threshold, tests)
X |
The tree of annotation terms as retrieved by the statistical analysis function (e.g. evalClusterHyper) |
threshold |
The selection threshold for the p-values. |
tests |
The number of annotation terms tested. |
The function will walk through all nodes of the tree specified as X and remove all annotation terms with a p-value lower than the given threshold divided by the number of tests (Bonferroni correction). The resulting tree will have the same structure as before but it can contain nodes that are empty.
A reduced tree with annotation elements having a significant p-value. This type of multiple testing correction assumes that the annotation terms are independant and should thus not be used for annotation data sets as for exmple the gene ontology.
Gunnar Wrobel, work@gunnarwrobel.de, http://www.gunnarwrobel.de.
set.seed(1000) data(benomylsetupsmall) a <- new("goCluster") execute(a) <- benomylsetupsmall ## The statistics are saved in the "statset" slot of ## the "stat" object that is itself a child of the "sign" ## object of a "goCluster" object ## We select all items with a p-value below 0.05 after ## Bonferroni correction. selectStatsBonferroni(a@sign@stat@statset,0.05,length(unlist(a@sign@stat@statset)))