NormiR: ExiMiR high-level function for miRNA raw data normalization.

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

View source: R/NormiR.R

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

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
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

1
2
3
4
5
6
data(galenv)
data(GSE20122)
eset.spike <- NormiR(GSE20122,
                     bg.correct=FALSE,
                     normalize.method='spikein',
                     summary.method='medianpolish')

ExiMiR documentation built on Nov. 8, 2020, 8:26 p.m.