loessQuantile.MAData: Estimates the (25%,50%,75%) quantile M(A) curves using loess
In HenrikBengtsson/aroma: An R Object-oriented Microarray Analysis package

Description Usage Arguments Author(s) See Also Examples

Estimates the (25%,50%,75%) quantile M(A) curves using loess. This method is used by *getAdjustedSpotVariability().

1 2	## S3 method for class 'MAData' loessQuantile(this, slides=NULL, subset=NULL, method=c("each", "average"), factor=1.4826, family="symmetric", ...)

`slides`	The slides which should be included in the calculations. If `NULL`, all slides are included.
`subset`	Indices of spots to be used for the estimates. Selecting a random subset may speed up the estimation and still give the same result. If `NULL`, all spots are included.
`method`	If `"each"`, estimates of the quantiles are given for each of the slides. If `"average"`, an estimate of the quantiles is given for all spots as if they came from the same slide.
`factor`	`numeric` factor to multiply to the upper and lower quantiles. If the data is Gaussian distributed the default factor `1.4826` makes the upper and lower quantile to correspond to one standard deviation.
`family`	Family to be used by `loess`. By default, a robust family is used.
`...`	Other arguments passed to `loess()`.

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

*getAdjustedSpotVariability(). 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)

   # Create four sets of slides where slide 2 and 4 are shifted R=G=a
   a <- 2^11
   ma <- list()
   ma[[1]] <- getSignal(raw, bgSubtract=TRUE)
   ma[[2]] <- clone(ma[[1]])
   shift(ma[[2]], R=a, G=a)
   ma[[3]] <- clone(ma[[1]])
   normalizeWithinSlide(ma[[3]], method="p")
   ma[[4]] <- clone(ma[[3]])
   shift(ma[[4]], R=a, G=a)

   setMethodS3("lines", "LoessQuantile", function(object, ...) {
     lines(object$x, object$centre, ...);
     lines(object$x, object$above, ...);
     lines(object$x, object$below, ...);
   })

   # Four plots
   subplots(4)
   Alim <- Mlim <- NA
   for (k in 1:length(ma)) {
     Alim <- range(c(Alim, ma[[k]]$A), na.rm=TRUE)
     Mlim <- range(c(Mlim, ma[[k]]$M), na.rm=TRUE)
   }
 
   for (k in 1:length(ma)) {
     plot(ma[[k]], slide=1, xlim=Alim, ylim=Mlim, col=k+1)
     fit <- loessQuantile(ma[[k]], slide=1)[[1]]
     lines(fit, lwd=2)
   }