%
% NOTE -- ONLY EDIT THE .Rnw FILE!!!  The .tex file is
% likely to be overwritten.
%


% \VignetteIndexEntry{Visualization of Microarray Data}
% \VignetteDepends{geneplotter, hgu95av2.db}
% \VignetteKeywords{Expression Analysis}
%\VignettePackage{geneplotter}
\documentclass[12pt]{article}

\usepackage{amsmath}
\usepackage{hyperref}

\newcommand{\Rfunction}[1]{{\texttt{#1}}}
\newcommand{\Robject}[1]{{\texttt{#1}}}
\newcommand{\Rpackage}[1]{{\textit{#1}}}

\author{Robert Gentleman}

\begin{document}
\title{Overview: Visualization of Microarray Data}
\maketitle{}

\section{Overview}

In this document we present a brief overview of the visualization
methods that are available in Bioconductor project.
To make use of these tools you will need the packages: \Rpackage{Biobase},
\Rpackage{annotate}, and \Rpackage{geneplotter}. These must be installed in your
version of R and when you start R you must load them with the
\Rfunction{library} command.

A quick word of warning regarding the interpretation of these
plots. We can only plot where the gene is supposed to be. If there are
translocations or amplifications these will not be detected by
microarray analyses.

<<getl>>=
library(geneplotter)
@


\section{Whole Genome Plots}

The functions \Rfunction{cPlot} and \Rfunction{cColor} allow the user to
associate microarray expression data with chromosomal location.
The plots can include any subset (by default all chromosomes are
shown) of chromosomes for the organism being considered.

To make these plots we use the complete reference set of genes
for the organism being studied. We must then obtain the chromosomal
location (in bases) and orientation (which strand) the gene is on. 
Chromosomes are represented by straight lines parallel to the
$x$--axis. Genes are represented by short perpendicular lines. All
genes for the experiment (i.e. for an Affymetrix U95A analysis we show
all genes on the chips). 

The user can then change the color of different sets of the
genes according to their needs.

The original setup is done using \Rfunction{cPlot}. The subsequent coloring
is done using \Rfunction{cColor}.


We will use the example data in \Robject{sample.ExpressionSet} to show how this function
might be used.

<<start>>=
data(sample.ExpressionSet)
eset = sample.ExpressionSet # legacy naming
mytt <- function(y, cov2) {
      ys <- split( y, cov2 )
      t.test( ys[[1]], ys[[2]] )
      }

ttout <- esApply(eset, 1, mytt, eset$type)
s1means <- sapply(ttout, function(x) x$estimate[1])
s2means <- sapply(ttout, function(x) x$estimate[2])
deciles <- quantile(c(s1means, s2means), probs=seq(0,1,.1))
s1class <- cut(s1means, deciles)
names(s1class) <- names(s1means)
s2class <- cut(s2means, deciles)
names(s2class) <- names(s2means)
@

Next we need to set up the graphics output. We do this in a rather
complicated way.
In the plot below we can compare the mean expression levels for genes
in Group 1 with those in Group 2.
The Group 1 values are in the left--hand plot and the Group 2 values
are in the right--hand plot.

\begin{verbatim}
cols <- dChip.colors(10)
nf <- layout(matrix(1:3,nr=1), widths=c(5,5,2))
chrObj <- buildChromLocation("hgu95av2")
cPlot(chrObj)
cColor(featureNames(eset), cols[s1class], chrObj)
cPlot(chrObj)
cColor(featureNames(eset), cols[s2class], chrObj)
image(1,1:10,matrix(1:10,nc=10),col=cols, axes=FALSE,
         xlab="", ylab="")
axis(2, at=(1:10), labels=levels(s1class), las=1)
\end{verbatim}


\begin{center}
<<f11,fig=TRUE, echo=FALSE>>=
cols <- dChip.colors(10)
def.par <- par(no.readonly = TRUE)# save default, for resetting...
nf <- layout(matrix(1:3,nr=1), widths=c(5,5,2))
chrObj <- buildChromLocation("hgu95av2")
cPlot(chrObj)
cColor(featureNames(eset), cols[s1class], chrObj)
cPlot(chrObj)
cColor(featureNames(eset), cols[s2class], chrObj)
image(1,1:10,matrix(1:10,nc=10),col=cols, axes=FALSE,
         xlab="", ylab="")
axis(2, at=(1:10), labels=levels(s1class), las=1)
par(def.par)
@
\end{center}

\section{Single Chromosome Plots}

A different view of the variation in expression level can be obtained
by plotting characteristics of expression levels over contiguous
regions of a chromosome.
For these plots cummulative expression or per--gene expressions can be
plotted. 
There are some issues of interpretation here (as in most places) --
expression is not likely to be controlled too much by chromosomal
locality. However these plots may be helpful in detecting
deletions (of both chromatids) or amplifications, or other interesting
features of the genome.

In this section we will show how one can explore a particular
chromosome for an amplicon.
The data arise from a study of breast cancer in the Iglehart
Laboratory.

\begin{center}
<<f22,fig=TRUE, echo=FALSE>>=
 par(mfrow=c(1,1)) 
 mycols <- c("red", "darkgreen", "blue")[eset$cov3]
 alongChrom(eset, "1", chrObj, plotFormat="cumulative", 
        col=mycols)
@
\end{center}

\end{document}