mdGsa: Multi-Dimensional Gene Set Analysis.
In mdgsa: Multi Dimensional Gene Set Analysis.

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

Performs a Multi-Variate Gene Set Analysis for two genomic measurements.

1
2
3

mdGsa(index, annot, p.adjust.method = "BY", family = quasibinomial(),
  verbose = TRUE, verbosity = 100, fulltable = FALSE,
  useColnames = TRUE, ...)

`index`	ranking index, generally a two column matrix.
`annot`	an annotation list.
`p.adjust.method`	p-value adjustment method for multiple testing.
`family`	see `glm`.
`verbose`	verbose.
`verbosity`	integer indicating which iterations should be indicated when verbose = TRUE.
`fulltable`	if TRUE, 'sd', 't' and 'convergence' indicator from the glm fit are included in the output.
`useColnames`	if TRUE the names of the two first columns of the matrix 'index' are used in the results data.frame.
`...`	further arguments to be pasted to `glm.fit`, for instance 'weights'.

'index' must be a numerical matrix or data.frame with at least two columns.

If there are more than three columns, the ranking indexes are taken form the two first one. The remaining columns are used as covariates to correct for within the analysis.

Default p-value correction is "BY".

In the output data.frame there are three parameters of each type: 'lor', 'pval', ... one for each of the two genomic conditions analyzed and the third one for the interaction between them.

If available, names of the fist two columns of the index matrix are used in the output data.frame. Changing the order of these two first columns will change the report order, but will not change the interpretation of the results. See Montaner et al. (2010) for further details on the algorithm.

A data.frame with a row for each Gene Set or block. Columns are:

N:: number of genes annotated to the Gene Set.
lor:: log Odds Ratio estimated for the Gene Set.
pval:: p-values associated to each log Odds Ratio.
padj:: adjusted p-values.
sd:: standard deviations associated to each log Odds Ratio.
t:: t statistic associated to each log Odds Ratio.

Apart from the 'N' coefficient, all other indices appear in triplicate: one coefficient for each genomic condition and a third one for the interaction.

David Montaner dmontaner@cipf.es

Montaner et al. (2010) "Multidimensional Gene Set Analysis of Genomic Data." PLoS ONE.

uvGsa, mdPat, glm.fit, p.adjust

rindexMat <- matrix (rnorm (2000), ncol = 2)
colnames (rindexMat) <- c ("genomicVar1", "genomicVar2")
rownames (rindexMat) <- paste0 ("gen", 1:1000)

annotList <- list (geneSet1 = sample (rownames (rindexMat), size = 10),
                   geneSet2 = sample (rownames (rindexMat), size = 15),
                   geneSet3 = sample (rownames (rindexMat), size = 20))

res <- mdGsa (rindexMat, annotList)
res