NormiR {ExiMiR} | R Documentation |
ExiMiR high-level function for miRNA raw data normalization.
Description
This function applies a standard raw data normalization pipeline (i.e. background
correction, normalization, PM correction if needed, and summarization) on the
input AffyBatch
object and returns the result in an ExpressionSet
object.
The methods supported by NormiR
for the background correction are provided by
the affy
or limma
packages, depending on whether the input AffyBatch
object has been created with ReadAffy
or ReadExi
/createAB
,
respectively.
By default, it applies the spike-in probe-based method for the second step of
normalization. In case the spike-in probe-based method cannot be applied, a median
normalization is executed instead. Several options allow however to force the execution
of the spike-in probe-based normalization and to fine-tune the resulting correction
functions.
The next step of PM correction is enabled only when numerical values are available
for the MM probes of the input AffyBatch
object. In this case the methods proposed
by NormiR
are provided by the affy
package.
The methods supported by NormiR
for the last step of summarization are also
provided by the affy
package. They do not depend on how the input AffyBatch
object has been created.
Usage
NormiR(abatch,
# background correction
bg.correct=TRUE,
bgcorrect.method='auto',
bgcorrect.param=list(),
# normalize
normalize=TRUE,
normalize.method='spikein',
normalize.param=list(),
# pm correction (enabled only when MM-values are available)
pmcorrect.method='pmonly',
pmcorrect.param=list(),
# expression values
summary.method='medianpolish',
summary.param=list(),
summary.subset=NULL,
# misc.
verbose=FALSE,
...)
Arguments
abatch |
An AffyBatch object.
|
bg.correct |
Logical. Default is TRUE : the background correction step will be performed.
|
bgcorrect.method |
Character vector. It contains the name of the background correction method. Running
NormiR.bgcorrect.methods(abatch) indicates which methods can be used, depending
on the raw data contained in the abatch object. The auto option
corresponds to the default choice of applying rma for single-channel arrays
and normexp for dual-channel arrays.
|
bgcorrect.param |
A R list containing the parameters required by the selected background correction method,
as specified in the documentation of the original functions bg.correct of the
affy package or backgroundCorrect of the limma package. As an illustration
the parameters of the normexp method of the limma package are given below.
- normexp.method
-
Character vector. The variant of the normexp method, matching exactly
the argument normexp.method of the backgroundCorrect function.
- offset
-
Numeric value to add to intensities. It matches exactly the argument
offset of the backgroundCorrect function.
|
normalize |
Logical. Default is TRUE : the normalization step will be performed.
|
normalize.method |
Character vector. It contains the name of normalization method. By default, the
spikein method is used. Running NormiR.normalize.methods(abatch) indicates
which other methods can be chosen, depending on the raw data contained in the
abatch object.
|
normalize.param |
A R list of the arguments that are used to control the spikein normalization. Running
NormiR.spikein.args() provides a complete list of all the tunable parameters supported by
NormiR and explained below.
- figures.output
-
Character vector. By default, display is used. Figures are shown to the screen. Using file
generates the figures in PDF format in the working directory.
- min.corr
-
Numeric. Default value is 0.5. Minimal allowed value for the average of the off-diagonal elements of the
Pearson correlation matrix of the spike-in probeset intensities across the arrays.
- loess.span
-
Numeric. Default value is -1, which corresponds to a loess smoothing neighbourhood spanning a fraction
3/(number of spike-in probesets) of the total number of points. Other positive values are allowed, see
the span argument of the R loess function
- extrap.points
-
Numeric. Default value is 2. The number of spike-in probesets used in the high-intensity extrapolation of
the normalization correction function.
- extrap.method
-
Character vector. Default value is mean . The method used for the high-intensity extrapolation of
the normalization correction function.
- force.zero
-
Logical. Default value is FALSE . If TRUE , it forces the normalization correction functions to
have zero values at the lower end of the probe intensity range.
- cover.ext
-
Numeric. Default value is 1/2. Minimal allowed relative coverage of the spike-in probesets intensities. It
is computed as the ratio between the intensity range covered by the spike-in probes and the one covered by
all probes on the array.
- cover.int
-
Numeric. Default value is 1/3. Maximal allowed relative intensity interval between two consecutive
spike-in probesets. It is computed as the largest intensity difference between two consecutive spike-in
probesets divided by the overall probe intensity range.
- verbose
-
Logical. Default is TRUE ; some details are provided on the console.
- max.log2span
-
Numeric. Default value is 1. Gives the maximal (log2) intensity interval allowed for the probes belonging
to one spike-in probeset.
- probeset.list
-
Vector of probes names that will be used as the "spike-in probes". By default, NormiR uses the
probes annotated as "spike-in" by Exiqon or Affymetrix.
|
pmcorrect.method |
Character vector. It contains the name of the PM correction method, which is enabled
only when numerical values are available for the MM probes of the input
AffyBatch object. Running NormiR.pmcorrect.methods(abatch) indicates
which other methods can be chosen instead of the default one pmonly .
|
pmcorrect.param |
A R list of optional parameters for the selected pmcorrect.method , as specified
in the documentation of the original function pmcorrect function of the affy package.
|
summary.method |
Character vector. It contains the name of the summarization method. Running
NormiR.summary.methods() indicates which other methods can be chosen instead
of the default one medianpolish .
|
summary.param |
A R list of optional parameters for the selected summary.method , as specified
in the documentation of the original AffyBatch method computeExprSet contained in
the affy package.
|
summary.subset |
A R list of probe set identifiers. When set to its default NULL value, the
summarized expression values are computed for all probe sets available on the array.
|
verbose |
Logical. The default value is TRUE . The details of the function execution are displayed on the console.
|
... |
Any additional argument. Used for backward compatibility.
|
Details
See accompanying vignette.
Value
An ExpressionSet
object containing the normalized expression data.
Author(s)
Sylvain Gubian, Alain Sewer, PMP SA
See Also
bg.correct.miR
,
NormiR.bgcorrect.methods
,
norm.miR
,
NormiR.normalize.methods
,
NormiR.spikein.args
,
NormiR.pmcorrect.methods
,
summarize.miR
,
NormiR.summary.methods
.
Examples
data(galenv)
data(GSE20122)
eset.spike <- NormiR(GSE20122,
bg.correct=FALSE,
normalize.method='spikein',
summary.method='medianpolish')
[Package
ExiMiR version 2.35.0
Index]