NormiR: ExiMiR high-level function for miRNA raw data normalization.
In ExiMiR: R functions for the normalization of Exiqon miRNA array data

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

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.

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,
       ...)

`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.

See accompanying vignette.

An ExpressionSet object containing the normalized expression data.

Sylvain Gubian, Alain Sewer, PMP SA

bg.correct.miR, NormiR.bgcorrect.methods, norm.miR, NormiR.normalize.methods, NormiR.spikein.args, NormiR.pmcorrect.methods, summarize.miR, NormiR.summary.methods.

data(galenv)
data(GSE20122)
eset.spike <- NormiR(GSE20122,
                     bg.correct=FALSE,
                     normalize.method='spikein',
                     summary.method='medianpolish')