normalizeAffineShift.MAData: Within-slide normalization that adjust the channel bias
In HenrikBengtsson/aroma: An R Object-oriented Microarray Analysis package

Description Usage Arguments Value Author(s) See Also Examples

Within-slide normalization that adjust the channel bias with the objective to minimize intensity dependency of the log-ratios.

1 2	## S3 method for class 'MAData' normalizeAffineShift(this, slides=NULL, groupBy=NULL, weights=NULL, method=c("Kerr", "Newton", "Bengtsson"), interval=c("safe", "auto"), ...)

`slides`	The slides which should be included in the calculations. If `NULL`, all slides are included.
`groupBy`	`character` string or `LayoutGroups` specifying the groups of spots that are to normalized individually. If `NULL`, all spots are normalized together.
`method`	`character` string specifying which method to use. If `"Kerr"`, perpendicular translation normalization is applied. If `"Newton"`, parallel translation normalization is applied. If `"Bengtsson"`, semi-perpendicular translation normalization is applied.
`interval`	The interval of possible shifts. Either a `numeric` `vector` of length 2 or a `character` string. If `"safe"`, the interval will be set, depending on method, such that no signals are made non-positive. If `"auto"`, the interval is choosen automatically depending on method. For some methods, `"auto"` gives the same result as `"safe"`.
`...`	Currently not used.

Returns a list structure containing information about the fit etc.

Henrik Bengtsson (http://www.braju.com/R/)

For more information see MAData.

SMA$loadData("mouse.data")
layout <- Layout$read("MouseArray.Layout.dat", path=system.file("data-ex", package="aroma"))
raw <- RawData(mouse.data, layout=layout)
methods <- c("Kerr", "Bengtsson")
Alim <- c(6,16)
Mlim <- c(-1,1)*3
subplots(2*length(methods))
for (method in methods) {
  ma <- getSignal(raw, bgSubtract=TRUE)
  fit <- normalizeAffineShift(ma, method=method, slide=1)
  a <- fit$shift[1,1]
  msg <- sprintf("Optimal shift according to '", method, "' is a = %1.f.\n", a)
  cat(msg)
  plot(ma, xlim=Alim, ylim=Mlim)
  lowessCurve(ma)
 
  x <- seq(-2*a,2*a, length=200)
  y <- sapply(x, FUN=fit$objectiveFunction, R=getR(ma),G=getG(ma))
  plot(x,y, pch=176, xlab="shift", ylab="objective");
  lines(x,y)
  abline(v=a)
} # for (method ...)