GOtree: GOtree, plot.GOtree and GOtreeWithLeaveOneOut

In pcaGoPromoter: pcaGoPromoter is used to analyze DNA micro array data

Description

GOtree finds significantly overrepresented Gene Ontology terms in a list of probes (only Biological processes) and return an object of type 'GOtree'.

print.GOtree list significant GO terms from an object of type 'GOtree'.

plot.GOtree creates a visual representation of the GO connection from an object of type 'GOtree'.

GOtreeHits return the genes/probes for a specific GO term.

GOtreeWithLeaveOut returns the same as GOtree, but run through the samples multiple times with 'Leave one out' cross-validation.

Usage

GOtree(input, inputType = "hgu133plus2", org = "Hs",
       statisticalTest = "binom", binomAlpha = NA,
       p.adjust.method = "fdr")

## S3 method for class 'GOtree'
print(x, ...)

## S3 method for class 'GOtree'
plot(x, boxes = 25, legendPosition = "topright",
     main = "Gene Ontology tree, biological processes", ...)

GOtreeHits(input, inputType = "hgu133plus2", org = "Hs",
           GOid, returnGeneSymbols = TRUE)

GOtreeWithLeaveOut(exprsData, inputType = "hgu133plus2", org = "Hs",
                   pc = 1, decreasing = TRUE, noProbes = 1000,
                   leaveOut = 1, runs = NCOL(exprsData))

Arguments

input	a character vector of Affymetrix probe ids, gene symbols or Entres gene IDs.
inputType	a character vector description the input type. Must be Affymetrix chip type, "geneSymbol" or "entrezID". The following Affymetrix chip type are supported: hgu133plus2, mouse4302, rat2302, hugene10st and mogene10st. Default is Affymetrix chip type "hgu133plus2".
org	a character vector with the organism. Can be "Hs", "Mm" or "Rn". Only needed if inputType is "geneSymbol" or "entrezID". See details. Default is "Hs".
statisticalTest	a character vector with the statistical method to be used. Can be "binom" or "fisher". Default is "binom".
binomAlpha	a value with the pvalue for use in self contained test.
p.adjust.method	the method for adjust p-values due to multiple testing. This will come in a separate column.
x	an object of type 'GOtree'.
boxes	an integer indication the amount of boxes (terms) in the plot.
legendPosition	a vector description the position of the legend. See ?xy.coords for possibilities. Set to NULL for no legend. Default is "topright".
main	a title for the GO tree plot
...	other parameters to be passed through to plotting functions.
GOid	a vector with the GO term of interest.
returnGeneSymbols	a logical indication whether gene symbols or probe ids should be returned. Default is gene symbols.
exprsData	A table with expression data. Row names should be probe identifiers (Affymetrix Probe set ID, Gene Symbols or Entrez gene ID). Column names should be sample identifiers.
pc	a number indication which principal component to extract the probe list based on the loading values from the pca.
decreasing	a logical value indication whether the loadings should be sorted in decreasing of ascending order ( decreasing == FALSE ). Decreasing order yields information about the positive direction and ascending order about the negative direction of the particular principal component.
noProbes	a number indicating the number of probes included in the calculations
leaveOut	a number indication what percentage to leave out in the cross-validation. If set to 1, each observation would be left out once and runs is set equal to number of observations. Deafault is 1.
runs	a number indicating how many times to run with leave out. If leaveOut = 1, runs is overrided with number of observations.

Details

GOtree returns a GOtree object. In contains a list of significant GO terms. plot() generated a visual plot of the GO tree.

GOtreeHits returns a vector with the genes/probes in a specific GO term.

GOtreeWithLeaveOut repeats function GOtree, but with different input. GOtreeWithLeaveOut takes a table of expression data as input, performs PCA, extracts probes / genes for the specified principal component and subsequenly performs GOtree. This is repeated the specified number of times. It can run with leave one out or with leave out a percentage. Only the GO terms that is found overrepresented in all the runs "qualifies" and a new p-value is calculated as the median of the p-value from all the runs. An object of type GOtree is returned.

Value

GOtree returns a object of type GOtree.

GOtreeWithLeaveOut returns a object of type GOtree.

Note

GOtreeWithLeaveOut may take some time to run - depending on the number of samples.

Author(s)

Morten Hansen [e-mail] and Jorgen Olsen [e-mail]

Examples

library(serumStimulation)
data(serumStimulation)
pcaOutput <- pca( serumStimulation )
posLoadings <- getRankedProbeIds( x=pcaOutput )
GOs <- GOtree( input=posLoadings[1:1000] )
GOs
plot(GOs, legendPosition=NULL)

## Not run: 
GOs <- GOtreeWithLeaveOut( exprsData=serumStimulation ) 
## End(Not run)

pcaGoPromoter documentation built on Oct. 31, 2019, 5:31 a.m.