normalizeWithinSlide.MAData: Within-slide normalization
In HenrikBengtsson/aroma: An R Object-oriented Microarray Analysis package

Description Usage Arguments Note Author(s) References See Also Examples

Performs a within-slide normalization slide by slide. For a detailed explanation of normalization, see [1].

Note that the data in the object is replaced with the new normalized data and the old data is removed. To keep the old data, make a copy of the object before normalizing by using clone(ma), see clone.Object and example below.

1 2	## S3 method for class 'MAData' normalizeWithinSlide(this, method, slides=NULL, weights=NULL, lowess=NULL, f=0.3, ...)

`method`	The normalization method to be used. Currently there are four different methods; `"m"` - median normalization, which sets the median of log intensity ratios to zero, `"l"` - global lowess normalization, `"p"` - print-tip group lowess normalization, and `"s"` - scaled print-tip group lowess normalization.
`weights`	Weights between zero and one, that is, in [0,1], of each data point specifying how much that data points will affect the normalization. A data point with weight zero (or `FALSE`) will not affect the normalization, but will be normalized. Currently only 0-1 (or `FALSE`-`TRUE`) weights are supported. Non-zero weights are treated as ones.
`lowess`	When doing global lowess normalization, `method="l"`, it is possible to specify the lowess line to be used. It is possible to specify individual lowess lines for each of the slides by letting lowess be a list of lines. If `lowess=NULL` the lowess curve will be estimated from the data.

Note that only one normalization is needed, i.e. doing different normalizations in serie on the same data set will not affect the results.

Also note that it is only the log ratios, M, are affected by the normalization, i.e. the log intensities, A, are not changed.

Note that the layout must be set for print-tip (method="p") and scaled (method="s") normalization. If layout is not set, an exception will be thrown. Normally, the layout is already set, such as when the data is read from for instance GenePix, ScanAlyze and Spot.

Henrik Bengtsson (http://www.braju.com/R/). Initial code for support of 'weights' by Jon McAuliffe, Statistics Dept, UC Berkeley. The original code was written by the sma authors Yee Hwa Yang yeehwa@stat.berkeley.edu Sandrine Dudoit sandrine@stat.berkeley.edu and Natalie Roberts nroberts@wehi.edu.au.

\item

[1]S. Dudoit, Y. H. Yang, M. J. Callow, and T. P. Speed. Statistical methods for identifying differentially expressed genes in replicated cDNA microarray experiments (Statistics, UC Berkeley, Tech Report 578). URL: http://www.stat.berkeley.edu/users/terry/zarray/Html/papersindex.html

For across-slide normalization see *normalizeAcrossSlides(). For more information see MAData.

   # The option 'dataset' is used to annotate plots.
   options(dataset="sma:MouseArray")

   SMA$loadData("mouse.data")
   layout <- Layout$read("MouseArray.Layout.dat", path=system.file("data-ex", package="aroma"))
   raw <- RawData(mouse.data, layout=layout)
   ma <- getSignal(raw)

   # Clone the data to get one non-normalized and one normalized data set.
   ma.norm <- clone(ma)

   # Normalize the data within slides using scaled print-tip normalization.
   normalizeWithinSlide(ma.norm, "s")

   # Plot data before and after normalization.
   layout(matrix(1:4, ncol=2, byrow=TRUE))
   plot(ma)
   plotSpatial(ma)
   plot(ma.norm)
   plotSpatial(ma.norm)