\section{Example phenotypes and interactions}\label{examplePhenotypes}
\subsection{Preliminaries}

<<exampleSingleKnockDownEffects1>>=
library("HD2013SGI")

data("Interactions",package="HD2013SGI")
data("mainEffects",package="HD2013SGI")

dir.create(file.path("result","Figures"),
           recursive=TRUE,showWarnings=FALSE)
@

\subsection{Examples of single knock down phenotypes}\label{exampleSingleKDPhenotypes}

Main effects were first summarized over all four siRNA design pairs and
divided by the median deviation to compute a $z$-score. Bars in the following barplots show the
mean of main effects calculated over the two replicates of the
screen. Circle symbols represent individual replicate
measurements. Main effects for five different phenotypes are plotted
for the two genes SETD8 and TCF20.
%
<<exampleSingleKnockDownEffects2,results=hide>>=
features = c("count", "nuc.0.s.area","nuc.0.m.eccentricity",
             "cell.Bact.m.eccentricity","cell.act.m.majoraxis")
ylim = c(-2.947124, 7.329518)
@
<<exampleSingleKnockDownEffects3,echo=FALSE,results=hide>>=
pdf(file=file.path("result","Figures","singleKnockDown-SETD8.pdf"),
    width=2.5,height=5)
<<exampleSingleKnockDownEffects4,results=hide>>=
HD2013SGI:::plotExampleSingleGeneEffects(gene="SETD8",features,mainEffects,ylim)
<<exampleSingleKnockDownEffects5,echo=FALSE,results=hide>>=
dev.off()
pdf(file=file.path("result","Figures","singleKnockDown-TCF20.pdf"),
    width=2.5,height=5)
<<exampleSingleKnockDownEffects6,results=hide>>=
HD2013SGI:::plotExampleSingleGeneEffects(gene="TCF20",features,mainEffects,ylim)
<<exampleSingleKnockDownEffects7,echo=FALSE,results=hide>>=
dev.off()
@
\begin{center}
\includegraphics[width=0.3\textwidth]{result/Figures/singleKnockDown-SETD8.pdf}
\includegraphics[width=0.3\textwidth]{result/Figures/singleKnockDown-TCF20.pdf}
\end{center}

\subsection{Example of genetic interactions}\label{exampleDoubleKDPhenotypes}

Barcharts for example interactions are plotted.
<<exampleInteractions1,echo=FALSE,results=hide>>=
pdf(file=file.path("result","Figures","interaction-PHF16-SETD8.pdf"),
    height=4,width=3)
@
<<exampleInteractions1,results=hide>>=
HD2013SGI:::plotExampleInteractions(feature="cell.act.m.majoraxis",
                                    target="PHF16",query="SETD8",
                                    Interactions, mainEffects)
@
<<exampleInteractions2,echo=FALSE,results=hide>>=
dev.off()
pdf(file=file.path("result","Figures","interaction-SIN3A-RUVBL1.pdf"),
    height=4,width=3)
@
<<exampleInteractions3,results=hide>>=
HD2013SGI:::plotExampleInteractions(feature="nuc.0.m.eccentricity",
                                    target="SIN3A",query="RUVBL1",
                                    Interactions, mainEffects)
@
<<exampleInteractions4,echo=FALSE,results=hide>>=
dev.off()
@

\begin{center}
\includegraphics[width=0.3\textwidth]{result/Figures/interaction-PHF16-SETD8.pdf}
\includegraphics[width=0.3\textwidth]{result/Figures/interaction-SIN3A-RUVBL1.pdf}
\end{center}

%-------------------------------------------------------------
\subsection{Overlap of interactions}\label{OverlapOfInteractions}
%-------------------------------------------------------------
To show the overlap of interactions called in different phenotypes,
interactions with an adjusted $p$-value of $\le0.01$ were considered.
%
<<overlapOfInteractions1>>=
features = c("nuc.0.s.area","nuc.0.m.majoraxis",
             "cell.Bact.m.eccentricity","cell.act.m.majoraxis")
SIG = Interactions$padj[,,,,features] <= 0.01
SIG[is.na(SIG)] = FALSE
@
The \Sexpr{length(dim(SIG))}-dimensional array \Robject{SIG} of Boolean values indicating the significant
interactions was reshaped to a 2-dimensional table with one column for
each feature.
<<overlapOfInteractions2>>=
dim(SIG) = c(prod(dim(SIG)[1:4]), dim(SIG)[5])
row.names(SIG) = sprintf("A%d",seq_len(nrow(SIG)))
colnames(SIG) = HD2013SGI:::humanReadableNames[features]
@
%
A contingency table was computed considering all gene pairs that interacted in
at leat one phenotype.
%
<<overlapOfInteractions3>>=
I = which(apply(SIG, 1, any))
Overlap = table(as.data.frame(SIG[I,]))
Overlap
save(Overlap, file=file.path("result","Figures","Overlap.rda"))
@
%
A Venn diagram was generated with a webtool
(http://bioinfogp.cnb.csic.es/tools/venny/index.html). Lists of
identifiers of the significant genes were written to be used with the webtool.
%
<<overlapOfInteractions4,eval=FALSE>>=
L = apply(SIG, 2, function(x) names(which(x)))
for (i in seq_along(L)) {
  writeLines(L[[i]],
    file.path("result", "Figures", sprintf("interactionlist%d.txt",i)))
}
@
%
In paper, a Venn diagram is shown. An alternative plot
where the region areas are proportional to the numbers, is a mosaic
plot as shown here.
%
<<overlapOfInteractions5,echo=FALSE,results=hide>>=
pdf(file.path("result","Figures","Mosaicplot.pdf"))
@
<<overlapOfInteractions6,echo=FALSE,results=hide>>=
mosaic(Overlap, shade = TRUE)
@
<<overlapOfInteractions7,echo=FALSE,results=hide>>=
dev.off()
@
%
\begin{center}
  \includegraphics[width=0.7\textwidth]{result/Figures/Mosaicplot.pdf}
\end{center}