\HeaderA{vsn}{Variance stabilization and calibration for microarray data.}{vsn}
\keyword{robust}{vsn}
\begin{Description}\relax
Robust estimation of variance-stabilizing and calibrating 
transformations for microarray data. This is the main function of
this package; see also the vignette vsn.pdf.
\end{Description}
\begin{Usage}
\begin{verbatim}
vsn(intensities,
    lts.quantile = 0.5,
    verbose      = interactive(),
    niter        = 10,
    cvg.check    = NULL,
    describe.preprocessing = TRUE,
    subsample,
    pstart,
    strata)
\end{verbatim}
\end{Usage}
\begin{Arguments}
\begin{ldescription}
\item[\code{intensities}] An object that contains intensity values from
a microarray experiment. See
\code{\LinkA{getIntensityMatrix}{getIntensityMatrix}} for details.
The intensities are assumed to be the raw
scanner data, summarized over the spots by an image analysis program,
and possibly "background subtracted".
The intensities must not be logarithmically or otherwise transformed,
and not thresholded or "floored". NAs are not accepted.
See details.
\item[\code{lts.quantile}] Numeric. The quantile that is used for the resistant
least trimmed sum of squares regression. Allowed values are between
0.5 and 1. A value of 1 corresponds to ordinary least sum of squares
regression.
\item[\code{verbose}] Logical. If TRUE, some messages are printed.
\item[\code{niter}] Integer. The number of iterations to be used in the least
trimmed sum of squares regression.
\item[\code{cvg.check}] List. If non-NULL, this allows finer control of the
iterative least trimmed sum of squares regression. See details.
\item[\code{pstart}] Array. If not missing, user can specify start values
for the iterative parameter estimation algorithm. See 
\code{\LinkA{vsnh}{vsnh}} for details.
\item[\code{describe.preprocessing}] Logical. If TRUE, calibration and
transformation parameters, plus some other information are stored in
the \code{preprocessing} slot of the returned object. See details.
\item[\code{subsample}] Integer. If specified, the model parameters are
estimated from a subsample of the data only, the transformation is
then applied to all data. This can be useful for performance reasons.
\item[\code{strata}] Integer vector. Its length must be the same as nrow(intensities).
This parameter allows for the calibration and error model parameters
to be stratified within each array, e.g to take into account probe 
sequence properties, print-tip or plate effects.  
If \code{strata} is not specified, one pair of parameters is fitted
for every sample (i.e. for every column of \code{intensities}). If
\code{strata} is specified, a pair of parameters is fitted for every 
stratum within every sample. The strata are coded for by the different
integer values. The integer vector \code{strata} can be obtained
from a factor \code{fac} through \code{as.integer(fac)}, from
a character vector \code{str} through \code{as.integer(factor(fac))}.
\end{ldescription}
\end{Arguments}
\begin{Details}\relax
\bold{Overview:} 
The function calibrates for sample-to-sample variations through
shifting and scaling, and transforms the intensities to a scale where
the variance is approximately independent of the mean intensity.
The variance stabilizing transformation is equivalent to the
natural logarithm in the high-intensity range, and to a
linear transformation in the low-intensity range. In an intermediate
range, the \emph{arsinh} function interpolates smoothly between the
two. For details on the transformation, please see the help page for
\code{\LinkA{vsnh}{vsnh}}. The parameters are estimated through
a robust variant of maximum likelihood. This assumes that for
the majority of genes the expression levels are not much different
across the samples, i.e., that only a minority of genes (less than
a fraction \code{1-lts.quantile}) is differentially expressed.

Even if most genes on an array are differentially expressed, it may still
be possible to use the estimator: if a set of non-differentially expressed
genes is known, e.g. because they are external controls or reliable
'house-keeping genes', the transformation parameters can be fitted with
\code{vsn} from the data of these genes, then the transformation can be
applied to all data with \code{\LinkA{vsnh}{vsnh}}.

\bold{Format:} The format of the matrix of intensities is as follows:
for the \bold{two-color printed array technology}, each row
corresponds to one spot, and the columns to the different arrays
and wave-lengths (usually red and green, but could be any number).
For example, if there are 10 arrays, the matrix would have 20 columns,
columns 1...10 containing the green intensities, and 11...20 the
red ones. In fact, the ordering of the columns does not matter to
\code{vsn}, but it is your responsibility to keep track of it for
subsequent analyses.
For \bold{one-color arrays}, each row corresponds to a probe, and each
column to an array.

\bold{Performance:} This function is slow. That is due to the nested
iteration loops of the numerical optimization of the likelihood function
and the heuristic that identifies the non-outlying data points in the
least trimmed squares regression. For large arrays with many tens of
thousands of probes, you may want to consider random subsetting: that is,
only use a subset of the e.g. 10-20,000 rows of the data matrix
\code{intensities} to fit the parameters, then apply the transformation
to all the data, using \code{\LinkA{vsnh}{vsnh}}. An example for this can be
seen in the function \code{\LinkA{normalize.AffyBatch.vsn}{normalize.AffyBatch.vsn}}, whose code
you can inspect by typing \code{normalize.AffyBatch.vsn} on the R
command line.

\bold{Iteration control:} 
By default, if \code{cvg.check} is \code{NULL}, the function will run
the fixed number \code{niter} of iterations in the least trimmed sum
of squares regression. More fine-grained control can be obtained by
passing a list with elements \code{eps} and \code{n}. If the maximum
change between transformed data values is smaller than \code{eps} for
\code{n} subsequent iterations, then the iteration terminates.

\bold{Estimated transformation parameters:} 
If \code{describe.preprocessing} is \code{TRUE}, the transformation
parameters are returned in the \code{preprocessing} slot of the
\code{description} slot of the resulting 
\code{\LinkA{exprSet}{exprSet}} object, in the form 
of a \code{\LinkA{list}{list}} with three elements
\Itemize{
\item \code{vsnParams}: the parameter array (see \code{\LinkA{vsnh}{vsnh}} 
for details) 
\item \code{vsnParamsIter}: an array with dimensions 
\code{c(dim(vsnParams, niter))} that contains the parameter 
trajectory during the iterative fit process (see also 
\code{\LinkA{vsnPlotPar}{vsnPlotPar}}).
\item \code{vsnTrimSelection}: a logical vector that for
each row of the intensities matrix reports whether it was below
(TRUE) or above (FALSE) the trimming threshold.
}
If \code{intensities} has class \code{\LinkA{exprSet}{exprSet}},
and its \code{description} slot has class
\code{\LinkA{MIAME}{MIAME}}, then this list is appended to any
existing entries in the \code{preprocessing} slot. Otherwise, the
\code{description} object and its \code{preprocessing} slot are created.
\end{Details}
\begin{Value}
An object of class \code{\LinkA{exprSet}{exprSet}}.
Differences between the columns of the transformed intensities are 
"generalized log-ratios", which are shrinkage estimators of the natural
logarithm of the fold change. For the transformation parameters,
please see the Details.
\end{Value}
\begin{Author}\relax
Wolfgang Huber \url{http://www.ebi.ac.uk/huber}
\end{Author}
\begin{References}\relax
Variance stabilization applied to microarray data
calibration and to the quantification of differential expression,
Wolfgang Huber, Anja von Heydebreck, Holger Sueltmann, Annemarie
Poustka, Martin Vingron; Bioinformatics (2002) 18 Suppl.1 S96-S104.

Parameter estimation for the calibration and variance stabilization 
of microarray data, 
Wolfgang Huber, Anja von Heydebreck, Holger Sueltmann, 
Annemarie Poustka, and Martin Vingron;  
Statistical Applications in Genetics and Molecular Biology (2003)
Vol. 2 No. 1, Article 3.
http://www.bepress.com/sagmb/vol2/iss1/art3.
\end{References}
\begin{SeeAlso}\relax
\code{\LinkA{vsnh}{vsnh}}, \code{\LinkA{vsnPlotPar}{vsnPlotPar}}, 
\code{\LinkA{exprSet-class}{exprSet.Rdash.class}}, 
\code{\LinkA{MIAME-class}{MIAME.Rdash.class}},
\code{\LinkA{normalize.AffyBatch.vsn}{normalize.AffyBatch.vsn}}
\end{SeeAlso}
\begin{Examples}
\begin{ExampleCode}
data(kidney)
log.na = function(x) log(ifelse(x>0, x, NA))

if(interactive()) {
  x11(width=9, height=4.5)
  par(mfrow=c(1,2))
}
plot(log.na(exprs(kidney)), pch=".", main="log-log")

vsnkid = vsn(kidney)   ## transform and calibrate
plot(exprs(vsnkid), pch=".", main="h-h")

if (interactive()) {
  x11(width=9, height=4)
  par(mfrow=c(1,3))
}

meanSdPlot(vsnkid)
vsnPlotPar(vsnkid, "factors")
vsnPlotPar(vsnkid, "offsets")

## this should always hold true
params = preproc(description(vsnkid))$vsnParams
stopifnot(all(vsnh(exprs(kidney), params) == exprs(vsnkid))) 
\end{ExampleCode}
\end{Examples}