GSEA.mean.t: Gene Set Enrichment Analysis using output from samroc
Description
Based on a list of gene sets, e.g. pathways, in terms Affymtrix identifiers, these
sets are ranked with respect to regulation as measured by an effect in a linear model using the SAM statistic. Typical
applications include two-group comparisons or simple linear regression to clinical variable or gene expression of a given gene.
if "absolute" value of the absolute value of the samroc test statistic is used. If "original" no transformation. "maxmean" not available.
two.side
if TRUE a two-sided test is performed. Currently only two-sided test when type = "original" and else one-sided
cutoff
Gene sets with the number of members not falling within the interval given by cutoff are excluded
restand
if TRUE a 'restandardization' following Efron and Tibshirani (2006) is performed
Value
A matrix with columns normal approximation p-values, mean statistic, median statistic, and if type = "original", also
Wilcoxon signed ranks statistic based p-value.
Details
Restandardization based on Efron and Tibshirani (2006) introduced. For normal approximation of the gene set statistic
both the mean of the statstic, or the variance (and likewise for the Wilcoxon statistic), are obtained from the permutation distribution included in the samroc.result object.
Note that this will account for the dependency between genes.
References
Tian, Lu and Greenberg, Steven A. and Kong, Sek Won and Altschuler, Josiah and Kohane, Isaac S. and Park, Peter J.
(2005) Discovering statistically significant pathways in expression profiling studies, PNAS Vol. 102, nr. 38, pp. 13544-13549
Bradley Efron and Robert Tibshirani (2006) On testing of the significance of sets of genes, Technical report, Stanford