normalise {arrayMagic} | R Documentation |
An object of class exprSetRG
is generated
which contains the resulting normalised data.
normalise(arrayDataObject, subtractBackground = FALSE, method = "vsn", subGroups = NULL, channelsSeparately = FALSE, hybridisationGroups, spotIdentifier, verbose = TRUE)
arrayDataObject |
object of class arrayData ;
required; default missing.
arrayDataObject must contain the raw data, i.e.
a three dimensional array
(spot x channel x hybridisation),
cf. the class methods getIntensities
and intensities<- ,
and must contain information
on the hybridisations,
cf. class methods getHybAttr and
{hybAttrList<-} .
If the argument subGroups (see below)
is specified it must also contain
information on the spots, cf. the class methods
getSpotAttr and
spotAttr<- .
Note: Weights are only used by loess-type normalisations.
|
subtractBackground |
logical; default: FALSE |
method |
character string; required; default: "vsn ";
possible values:
"none ", "vsn ", "quantile ",
"loess ", "loessScale ", "loessQuantile ".
Note: "quantile " and "loess* " normalisation
transforms the data to the (natural) logarithmic scale.
Note: Weights are only used by loess-type normalisations.
Note: "loessScale " and "loessQuantile " refer
to a loess normalisation followed by a
between slide normalisation, cf. the function
normalizeBetweenArrays of the
limma package.
Note: "quantile " and "loess* " normalisations
do call the corresponding methods of the limma package
and "vsn " uses the vsn package.
|
subGroups |
character string or NULL ;
required; default: NULL ;
subGroups allows to define subgroups
of spots of each hybridisation
which are normalised separately like a
print-tip normalisation.
The list must contain a column name refering to the
data.frame of getSpotAttr(arrayDataObject) .
In case of GenePix data you may for example
specify the column name "Block".
The column itself must contain integer values.
Note: In case of method == "vsn" ,
vsn is called with
the argument strata , which differs
to a separate normalisation of each subgroup
with vsn itself.
|
channelsSeparately |
logical; required; default: FALSE ;
If channelsSeparately ist set to TRUE
each channel is normalised separately. Only meaningful
for single channel normalisation methods like
"vsn " and
"quantile " normalisation
but not for ratio based
normalisation like "loess ". |
hybridisationGroups |
list of vectors of indexes or
the character string "slideBySlide";
optional; default: missing.
Each group of hybridisations
is normalised separately. If
missing all hybridisations are
taken as one group. Only meaningful for
normalisation methods like "vsn" and "quantile".
The indexes must refer to the third
dimension of getIntensities(arrayDataObject)
and have to contain all hybridisations.
|
spotIdentifier |
character string; optional; default missing.
spotIdentifier specifies the column
of getSpotAttr(arrayDataObject) which must
contain non-unique spot or gene identifiers.
The identifiers are used as names for the
resulting exprSetRG -object; cf. the function
geneNames of the exprSetRG-class . |
verbose |
logical; required; default: TRUE |
object of class exprSetRG
Andreas Buness <a.buness@dkfz.de>
exprSetRG-class
,
processArrayData
,
processArrayDataObject
intensities <- array(data=runif(720),dim=c(120,2,3)) dimnames(intensities) <- list(NULL, c("green","red"), NULL) hybAttr <- data.frame(Name=I(c("hx","hy","hz")), Index=c(1:3)) arrayDataObject <- new("arrayData", intensities=intensities, hybAttrList=list(red=hybAttr,green=hybAttr) ) exprSetRGObject <- normalise(arrayDataObject = arrayDataObject, subtractBackground = FALSE, method = "none", verbose = TRUE ) nRed <- exprs(getExprSetRed(exprSetRGObject)) nGreen <- exprs(getExprSetGreen(exprSetRGObject)) stopifnot( all.equal.numeric( nRed, intensities[,"red",] ) ) stopifnot( all.equal.numeric( nGreen, intensities[,"green",] ) )