%\VignetteIndexEntry{topGO}
%\VignetteDepends{ALL, hgu95av2.db, genefilter, xtable, multtest, Rgraphviz}
%\VignetteKeywords{topGO, GO, graph}
%\VignettePackage{topGO}

\documentclass[a4paper, oneside, 10pt]{article}

\usepackage[pdftex]{graphicx}
\usepackage{calc}
\usepackage{sectsty}
\usepackage{caption}
\usepackage{natbib}
\renewcommand{\captionfont}{\it\sffamily}
\renewcommand{\captionlabelfont}{\bf\sffamily}
\allsectionsfont{\sffamily}

% page style %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\usepackage[a4paper, left=25mm, right=20mm, top=20mm, bottom=25mm, nohead]{geometry}
\setlength{\parskip}{1.5ex}
\setlength{\parindent}{0cm}
\pagestyle{empty}


\usepackage{Sweave}
\SweaveOpts{prefix.string = tGO}


\title{\vspace*{-6ex} Gene set enrichment analysis with {\bf topGO}}
\author{Adrian Alexa,  J\"org Rahnenf\"uhrer}
\date{\today \\%
  \texttt{http://www.mpi-sb.mpg.de/$\sim$alexa}}

\begin{document}
\maketitle

%%\newpage
\tableofcontents
\newpage

<<echo = FALSE>>=
options(width = 95)
## we better keep the data in data frames as strings
options(stringsAsFactors = FALSE)
@ 

\section{Introduction}

The {\tt topGO} package is designed to facilitate semi-automated enrichment analysis for Gene Ontology (GO) terms. 
The process consists of input of normalised gene expression measurements, gene-wise correlation or differential 
expression analysis, enrichment analysis of GO terms, interpretation and visualisation of the results.

One of the main advantages of {\tt topGO} is the unified gene set testing framework it offers.
Besides providing an easy to use set of functions for performing GO enrichment analysis, it
also enables the user to easily implement new statistical tests or new algorithms that deal with
the GO graph structure. This unified framework also facilitates the comparison between different
GO enrichment methodologies. 

% The package provides a set of classes and methods for performing enrichment analysis
% of Gene Ontology categories. The user can also use the available framework in order to
% develop and test new enrichment methods that make use of the GO graph structure.

There are a number of test statistics and algorithms dealing with the GO graph structured 
ready to use in {\tt topGO}. Table~\ref{tabletopGO} presents the compatibility table between 
the test statistics and GO graph methods.


<<echo = FALSE>>=
x <- topGO:::.algoComp[, -8]
x <- x[, colSums(x) > 0]
yesPic <- "\\includegraphics[width=3mm]{green_ckmark.png}"
noPic <- "\\includegraphics[width=3mm]{red_ckmark.png}"
x[x == 1] <-  yesPic
x[x == "0"] <-  noPic
@ 

\begin{table}[!ht]
  \centering\resizebox{.5\linewidth}{!}{%
<<echo = FALSE, results = tex>>=
if(require(xtable))
  print(xtable(x, align =  c("l", rep("c", ncol(x)))), 
        type = "latex", sanitize.text.function = function(x) return(x),
        floating = FALSE)

@
}\caption{Algorithms currently supported by {\bf topGO}.}
\label{tabletopGO}
\end{table}

The {\sf elim} and {\sf weight} algorithms were introduced in~\cite{Alexa2006}. The default algorithm
used by the {\tt topGO} package is a mixture between the {\sf elim} and the {\sf weight} algorithms and
it will be referred as {\sf weight01}. The {\sf parentChild} algorithm was introduced by~\cite{Grossmann2007}.

We assume the user has a good understanding of GO, see~\cite{GOC01}, and is familiar with gene set enrichment 
tests. Also this document requires basic knowledge of R language.

The next section presents a quick tour into {\tt topGO} and is thought to be independent of the rest
of this manuscript. The remaining sections provide details on the functions used in the sample
section as well as showing more advance functionality implemented in the {\tt topGO} package.

\section{Instalation}

This section briefly describe the necessary to get {\tt topGO} running on your system. 
We assume that the user have the R program (see the R project at http://www.r-project.org) already
installed and its familiar with it. You will need to have R 2.10.0 or later to be able to install and
run {\tt topGO}.

The {\tt topGO} package is available from the Bioconductor repository at http://www.bioconductor.org
To be able to install the package one needs first to install the core Bioconductor packages. 
If you have already installed Bioconductor packages on your system then you can skip the two lines 
below.

<<eval = FALSE>>=
source("http://bioconductor.org/biocLite.R")
biocLite()
@ 

Once the core Bioconductor packages are installed, we can install the {\tt topGO} package by

<<eval = FALSE>>=
source("http://bioconductor.org/biocLite.R")
biocLite("topGO")
@ 

\clearpage

\section{Quick start guide}
\label{sampleSession}

This section describes a simple working session using {\tt topGO}. There are only a handful of commands
necessary to perform a gene set enrichment analysis which will be briefly presented below.

A typical session can be divided into three steps:
\begin{enumerate}
\item {\it Data preparation:} List of genes identifiers, gene scores, list of differentially expressed genes
  or a criteria for selecting genes based on their scores, as well as gene-to-GO annotations are all collected
  and stored in a single {\tt R} object.
  
\item {\it Running the enrichment tests:} Using the object created in the first step the user can perform
  enrichment analysis using any feasible mixture of statistical tests and methods that deal with the GO topology.
  
\item {\it Analysis of the results:} The results obtained in the second step are analysed using 
  summary functions and visualisation tools.
\end{enumerate}

Before going through each of those steps the user needs to decide which biological question he would
like to investigate. The aim of the study, as well as the nature of the available data, will dictate which test 
statistic/methods need to used. 

In this section we will test the enrichment of GO terms with differentially expressed genes using two 
statistical tests, namely Kolmogorov-Smirnov test and Fisher's exact test.


\subsection{Data preparation}
In the first step a convenient {\tt R} object of class {\tt topGOdata} is created containing all the
information required for the remaining two steps. The user needs to provide the gene universe, GO annotations
and either a criteria for selecting interesting genes (e.g. differentially expressed genes) from the gene
universe or a score associated with each gene.

In this session we will test the enrichment of GO terms with differentially expressed genes. Thus, the 
starting point is a list of genes and the respective $p$-values for differential expression. A toy
example of a list of gene $p$-values is provided by the {\tt geneList} object.

<<results = hide>>=
library(topGO)
library(ALL)
data(ALL)
data(geneList)
@ 

The {\tt geneList} data is based on a differential expression analysis of the ALL(Acute Lymphoblastic Leukemia)
dataset that was extensively studied in the literature on microarray analysis~\cite{Chiaretti04}.
Our toy example contains just a small amount, $\Sexpr{length(geneList)}$, of genes and their corresponding $p$-values.
The next data one needs are the gene groups itself, the GO terms in our case, and the mapping that associate 
each gene with one or more GO term(s). The information on where to find the GO annotations is stored in the {\tt ALL}
object and it is easily accessible.

<<>>=
affyLib <- paste(annotation(ALL), "db", sep = ".")
library(package = affyLib, character.only = TRUE)
@ 

The microarray used in the experiment is the {\tt \Sexpr{annotation(ALL)}} from Affymetrix, as we can see 
from the {\tt affyLib} object. When we loaded the {\tt geneList} object a selection function used for 
defining the list of differentially expressed genes is also loaded under the name of {\tt topDiffGenes}.
The function assumes that the provided argument is a named vector of $p$-values. With the help of this
function we can see that there are \Sexpr{sum(topDiffGenes(geneList))} genes with a raw $p$-value less
than $0.01$ out of a total of \Sexpr{length(geneList)} genes.

<<>>=
sum(topDiffGenes(geneList))
@ 

We now have all data necessary to build an object of type {\tt topGOdata}. This object will
contain all gene identifiers and their scores, the GO annotations, the GO hierarchical structure
and all other information needed to perform the desired enrichment analysis. 

<<results = hide>>=
sampleGOdata <- new("topGOdata", 
                    description = "Simple session", ontology = "BP",
                    allGenes = geneList, geneSel = topDiffGenes,
                    nodeSize = 10,
                    annot = annFUN.db, affyLib = affyLib)
@ 

The names of the arguments used for building the {\tt topGOdata} object should be self-explanatory. 
We quickly mention that {\tt nodeSize = 10} is used to prune the GO hierarchy from the terms
which have less than $10$ annotated genes and that {\tt annFUN.db} function is used to extract the 
gene-to-GO mappings from the {\tt affyLib} object. Section~\ref{subsec:topGOdata} describes the
parameters used to build the {\tt topGOdata} in details.

A summary of the {\tt sampleGOdata} object can be seen by typing the object name at the R prompt.
Having all the data stored into this object facilitates the access to identifiers, annotations and 
to basic data statistics.

<<results = hide>>=
sampleGOdata
@ 


\subsection{Performing the enrichment tests}
Once we have an object of class {\tt topGOdata} we can start with the enrichment analysis. 
We will use  two types of test statistics: Fisher's exact test which is based on gene counts,
and a Kolmogorov-Smirnov like test which computes enrichment based on gene scores. We can use 
both these tests since each gene has a score (representing how differentially expressed a gene is)
and by the means of {\tt topDiffGenes} functions the genes are categorized into differentially 
expressed or not differentially expressed genes. All these are stored into {\tt sampleGOdata} object.

The function {\tt runTest} is used to apply the specified test statistic and method to the data.
It has three main arguments. The first argument needs to be an object of class {\tt topGOdata}.
The second and third argument are of type character; they specify the method for dealing with
the GO graph structure and the test statistic, respectively.

First, we perform a classical enrichment analysis by testing the over-representation of GO terms 
within the group of differentially expressed genes. For the method {\sf classic} each GO category 
is tested independently. 

<<results = hide>>=
resultFisher <- runTest(sampleGOdata, algorithm = "classic", statistic = "fisher")
@ 

{\tt runTest} returns an object of class {\tt topGOresult}. A short summary of this object is shown below.

<<>>=
resultFisher
@ 

Next we will test the enrichment using the Kolmogorov-Smirnov test. We will use the both the
{\sf classic} and the {\sf elim} method.

<<results = hide>>=
resultKS <- runTest(sampleGOdata, algorithm = "classic", statistic = "ks")
resultKS.elim <- runTest(sampleGOdata, algorithm = "elim", statistic = "ks")
@ 

Please note that not all statistical tests work with every method. The compatibility matrix between
the methods and statistical tests is shown in Table~\ref{tabletopGO}.

The $p$-values computed by the {\tt runTest} function are unadjusted for multiple testing. We do not
advocate against adjusting the $p$-values of the tested groups, however in many cases adjusted $p$-values
might be misleading.


\subsection{Analysis of results}
After the enrichment tests are performed the researcher needs tools for analysing and interpreting the
results. {\tt GenTable} is an easy to use function for analysing the most significant GO terms and the
corresponding $p$-values. In the following example, we list the top $10$ significant GO terms identified 
by the {\sf elim} method. At the same time we also compare the ranks and the $p$-values of these GO terms
with the ones obtained by the {\sf classic} method.

<<>>=
allRes <- GenTable(sampleGOdata, classicFisher = resultFisher, 
                   classicKS = resultKS, elimKS = resultKS.elim,
                   orderBy = "elimKS", ranksOf = "classicFisher", topNodes = 10)
@ 

The {\tt GenTable} function returns a data frame containing the top {\tt topNodes} GO terms identified
by the {\sf elim} algorithm, see {\tt orderBy} argument. The data frame includes some statistics on the
GO terms and the $p$-values corresponding to each of the {\tt topGOresult} object specified as arguments.
Table~\ref{tab:sampleGOresults} shows the results.

\begin{table}[!t]
  \centering\resizebox{.99\linewidth}{!}{%
<<echo = FALSE, results = tex>>=
if(require(xtable))
  print(xtable(apply(allRes, 2, as.character)), floating = FALSE)
@
}\caption{Significance of GO terms according to {\sf classic} and {\sf elim} methods.}
\label{tab:sampleGOresults}
\end{table}


For accessing the GO term's $p$-values from a {\tt topGOresult} object the user should use the {\tt score}
functions. As a simple example, we look at the differences between the results of the {\sf classic} and
the {\sf elim} methods in the case of the Kolmogorov-Smirnov test. The {\tt elim} method was design to be 
more conservative then the {\tt classic} method and therefore one expects the $p$-values returned by the
former method are lower bounded by the $p$-values returned by the later method. 
The easiest way to visualize this property is to scatter plot the two sets of $p$-values against each other. 


<<echo = FALSE>>=
colMap <- function(x) {
  .col <- rep(rev(heat.colors(length(unique(x)))), time = table(x))
  return(.col[match(1:length(x), order(x))])
}
@ 

<<eval = FALSE>>=
pValue.classic <- score(resultKS)
pValue.elim <- score(resultKS.elim)[names(pValue.classic)]

gstat <- termStat(sampleGOdata, names(pValue.classic))
gSize <- gstat$Annotated / max(gstat$Annotated) * 4
gCol <- colMap(gstat$Significant)

plot(pValue.classic, pValue.elim, xlab = "p-value classic", ylab = "p-value elim",
     pch = 19, cex = gSize, col = gCol)
@ 


\setkeys{Gin}{width=.9\linewidth}
\begin{figure}[!h]
  \centering 
<<fig = TRUE, echo = FALSE, width = 11>>=
pValue.classic <- score(resultKS)
pValue.elim <- score(resultKS.elim)[names(pValue.classic)]

gstat <- termStat(sampleGOdata, names(pValue.classic))
gSize <- gstat$Annotated / max(gstat$Annotated) * 4
gCol <- colMap(gstat$Significant)

par(mfcol = c(1, 2), cex = 1)
plot(pValue.classic, pValue.elim, xlab = "p-value classic", ylab = "p-value elim",
     pch = 19, cex = gSize, col = gCol)

plot(pValue.classic, pValue.elim, log = "xy", xlab = "log(p-value) classic", ylab = "log(p-value) elim",
     pch = 19, cex = gSize, col = gCol)

@ 
\caption{$p$-values scatter plot for the {\sf classic} ($x$ axis) and {\sf elim} ($y$ axis)
  methods. On the right panel the $p$-values are plotted on a linear scale. The left panned plots the same
  $p$-values on a logarithmic scale. The size of the dot is proportional with the number of annotated genes
  for the respective GO term and its coloring represents the number of significantly differentially 
  expressed genes, with the dark red points having more genes then the yellow ones.}
\label{scatterClassicElim}
\end{figure}

We can see in Figure~\ref{scatterClassicElim} that there are indeed differences between
the two methods. Some GO terms found significant by the {\sf classic} method are less 
significant in the {\sf elim}, as expected. However, we can visible identify a few 
GO terms for which the {\sf elim} $p$-value is less conservative then the {\sf classic} $p$-value.
We can identify these terms and find the number of annotated genes:


<<>>=
sel.go <- names(pValue.classic)[pValue.elim < pValue.classic]
cbind(termStat(sampleGOdata, sel.go),
      elim = pValue.elim[sel.go],
      classic = pValue.classic[sel.go])
@ 

It is quite interesting that such cases appear. These $\Sexpr{length(sel.go)}$ GO terms are rather
general (having many annotated genes) and their $p$-values are not significant at the $0.05$ level. 
Also the difference in the significance is rather small, bar one term. Therefore these GO terms 
would not appear in the list of top significant terms. More significant GO terms are less likely 
to be influenced by this non monotonic behavior.


Another insightful way of looking at the results of the analysis is to investigate how
the significant GO terms are distributed over the GO graph.
Figure~\ref{fig:sampleGOelim} shows the the subgraph induced by the $5$ most significant GO terms
as identified by the {\tt elim} algorithm. Significant nodes are represented as rectangles. 
The plotted graph is the upper induced graph generated by these significant nodes. 

<<eval = FALSE>>=
showSigOfNodes(sampleGOdata, score(resultKS.elim), firstSigNodes = 5, useInfo = 'all')
@ 

<<results = hide, echo = FALSE>>=
printGraph(sampleGOdata, resultKS.elim, firstSigNodes = 5, fn.prefix = "tGO", useInfo = "all", pdfSW = TRUE)
@ 

\begin{figure}[!ht]
\centering 
\includegraphics[width=1.05\linewidth]{tGO_elim_5_all}
\caption{The subgraph induced by the top $5$ GO terms identified by the {\sf elim} algorithm
  for scoring GO terms for enrichment. Rectangles indicate the $5$ most significant terms.
  Rectangle color represents the relative significance, ranging from dark red (most significant)
  to bright yellow (least significant). 
  For each node, some basic information is displayed. The first two lines show the  GO identifier
  and a trimmed GO name. In the third line the raw p-value is shown. The forth line is showing the 
  number of significant genes and the total number of genes annotated to the respective GO term. 
}
\label{fig:sampleGOelim}
\end{figure}
\clearpage



\section{Loading genes and annotations data}
\label{sec:topGOdata}


\subsection{Getting started}
\label{subsec:preprocessing}
To demonstrate the package functionality we will use the ALL(Acute Lymphoblastic Leukemia) gene expression
data from \cite{Chiaretti04}. The dataset consists of $128$ microarrays from different patients with ALL
measured using the HGU95aV2 Affymetrix chip.
Additionally, custom annotations and artificial datasets will be used to demonstrate specific features. 

We first load the required libraries and data:
<<results = hide>>=
library(topGO)
library(ALL)
data(ALL)
@ 

When the {\tt topGO} package is loaded three environments {\tt GOBPTerm, GOMFTerm} and {\tt GOCCTerm}
are created and bound to the package environment. These environments are build based on the {\tt GOTERM} 
environment from package {\tt GO.db}. They are used for fast recovering of the information specific to each
of the three ontologies: BP, MF and CC. In order to access all GO groups that belong to a specific ontology,
e.g. Biological Process (BP), one can type:
<<>>= 
BPterms <- ls(GOBPTerm)
head(BPterms)
@ 


Usually one needs to remove probes/genes with low expression value as well as probes with very small
variability across samples. Package {\tt genefilter} provides tools for filtering genes. In this 
analysis we choose to filter as many genes as possible for computational reasons;
working with a smaller gene universe allows us to exemplify more of the functionalities implemented
in the {\tt topGO} package and at the same time allows this document to be compiled in a
relatively short time.
The effect of gene filtering is discussed in more details in Section~\ref{subsec:GOannotations}.

%%selProbes <- genefilter(ALL, filterfun(pOverA(0.25, log2(100)), function(x) (IQR(x) > 0.5)))
<<>>=
library(genefilter)
selProbes <- genefilter(ALL, filterfun(pOverA(0.20, log2(100)), function(x) (IQR(x) > 0.25)))
eset <- ALL[selProbes, ]
@ 
The filter selects only $\Sexpr{nrow(eset)}$ probesets out of $\Sexpr{nrow(ALL)}$ probesets available 
on the $\Sexpr{annotation(ALL)}$ array.

\paragraph{The gene universe and the set of  interesting genes} \ \\
The set of all genes from the array/study will be referred from now on as the gene universe.
Having the gene universe, the user can define a list of interesting genes or to compute
gene-wise scores that quantify the significance of each gene. When gene-wise scores are available 
the list of interesting genes is defined to be the set of gene with a {\it significant} score. 
The {\tt topGO} package deals with these two cases in a unified way once the main data container, 
the {\tt topGOdata} object, is constructed. The only time the user needs to distinguish between 
these two cases is during the construction of the data container. 

Usually, the gene universe is defined as all feasible genes measured by the microarray. In the case of
the ALL dataset we have $\Sexpr{length(featureNames(eset))}$ feasible genes, the ones that were not
removed by the filtering procedure. 


\subsection{The {\tt topGOdata} object}
\label{subsec:topGOdata}
The central step in using the {\tt topGO} package is to create a {\tt topGOdata} object.
This object will contain all information necessary for the GO analysis, namely the list of genes,
the list of interesting genes, the gene scores (if available) and the part of the GO ontology
(the GO graph) which needs to be used in the analysis. 
The {\sf topGOdata} object will be the input of the testing procedures, the evaluation and visualisation 
functions.

To build such an object the user needs the following:
\begin{itemize}
\item A list of gene identifiers and optionally the gene-wise scores. The score can be the
  $t$-test statistic (or the $p$-value) for differential expression, correlation with a phenotype, 
  or any other relevant score.

\item A mapping between gene identifiers and GO terms. In most cases this mapping is directly available
  in Bioconductor as a microarray specific annotation package. In this case the user just needs to
  specify the name of the annotation to be used. For example, the annotation package needed for the ALL
  dataset is {\tt \Sexpr{affyLib}}.
  
  Of course, Bioconductor does not include up-to-date annotation packages for all platforms. 
  Users who work with custom arrays or wish to use a specific mapping between genes and GO terms, 
  have the possibility to load custom annotations. This is described in Section~\ref{subsec:annotations}.

\item The GO hierarchical structure. This structure is obtained from the {\tt GO.db} package. At the moment
  {\tt topGO} supports only the ontology definition provided by {\tt GO.db}.

\end{itemize}

We further describe the arguments of the {\tt initialize} function ({\tt new}) used to construct an instance
of this {\it data container} object.

\begin{description}
\item[{\tt ontology:}] character string specifying the ontology of interest (BP, MF or CC)
  
\item[{\tt description:}] character string containing a short description of the study [optional].
  
\item[{\tt allGenes:}] named vector of type numeric or factor. The names attribute contains the genes
  identifiers. The genes listed in this object define the gene universe.
  
\item[{\tt geneSelectionFun:}] function to specify which genes are interesting based on the gene scores. 
  It should be present iff the {\tt allGenes} object is of type numeric.

\item[{\tt nodeSize:}] an integer larger or equal to $1$. This parameter is used to prune the GO hierarchy
  from the terms which have less than {\tt nodeSize} annotated genes (after the true path rule is applied).
  
\item[{\tt annotationFun:}] function which maps genes identifiers to GO terms. There are a couple of annotation
  function included in the package trying to address the user's needs. The annotation functions take three
  arguments. One of those arguments is specifying where the mappings can be found, and needs to be provided
  by the user. Here we give a short description of each:
  \begin{description}
  \item[{\tt annFUN.db}] this function is intended to be used as long as the chip used by the user has an
    annotation package available in Bioconductor.
    
  \item[{\tt annFUN.org}] this function is using the mappings from the "org.XX.XX" annotation packages.
    Currently, the function supports the following gene identifiers: Entrez, GenBank, Alias, Ensembl,
    Gene Symbol, GeneName and UniGene. 
    
  \item[{\tt annFUN.gene2GO}] this function is used when the annotations are provided as a gene-to-GOs mapping.
    
  \item[{\tt annFUN.GO2gene}] this function is used when the annotations are provided as a GO-to-genes mapping.
    
  \item[{\tt annFUN.file}] this function will read the annotationsof the type gene2GO or GO2genes
    from a text file.    
  \end{description}

\item[{\tt ...:}] list of arguments to be passed to the {\tt annotationFun}
\end{description}



\subsection{Custom annotations}
\label{subsec:annotations}
This section describes how custom GO annotations can be used for building a {\tt topGOdata} object.

Annotations need to be provided either as {\it gene-to-GOs} or as {\it GO-to-genes} mappings.
An example of such mapping can be found in the "topGO/examples" directory. The file
"geneid2go.map" contains gene-to-GOs mappings.  For each gene identifier are listed the GO terms to
which this gene is specifically annotated. We use the {\tt readMappings} function to parse this file.

<<>>=
geneID2GO <- readMappings(file = system.file("examples/geneid2go.map", package = "topGO"))
str(head(geneID2GO))
@ 

The object returned by {\tt readMappings} is a named list of character vectors. The list names give the
genes identifiers. Each element of the list is a character vector and contains the GO identifiers
annotated to the specific gene. It is sufficient for the mapping to contain only the most specific GO
annotations. However, {\tt topGO} can also take as an input files in which all or some ancestors of the
most specific GO annotations are included. This redundancy is not making for a faster running time and
if possible it should be avoided.

The user can read the annotations from text files or they can build an object such as {\tt geneID2GO}
directly into R. The text file format required by the {\tt readMappings} function is very simple. It
consists of one line for each gene with the following syntax:

\begin{verbatim}
gene_ID<TAB>GO_ID1, GO_ID2, GO_ID3, ....
\end{verbatim}

Reading GO-to-genes mappings from a file is also possible using the {\tt readMappings} function.
However, it is the user responsibility to know the direction of the mappings. The user can easily
transform a mapping from gene-to-GOs to GO-to-genes (or vice-versa) using the function {\tt inverseList}:

<<>>=
GO2geneID <- inverseList(geneID2GO)
str(head(GO2geneID))
@ 


\subsection{Predefined list of interesting genes}  
\label{subsec:sigGenes}
If the user has some a priori knowledge about a set of interesting genes, he can test the enrichment of
GO terms with regard to this list of interesting genes. In this scenario, when only a list of interesting
genes is provided, the user can use only tests statistics that are based on gene counts, like Fisher's exact
test, Z score and alike.

To demonstrate how custom annotation can be used this section is based on the toy dataset, the {\tt geneID2GO}
data, from Section~\ref{subsec:annotations}. The gene universe in this case is given by the list names:

<<results = hide>>=
geneNames <- names(geneID2GO)
head(geneNames)
@ 

Since for the available genes we do not have any measurement and thus no criteria to select interesting genes,
we randomly select $10\%$ genes from the gene universe and consider them as interesting genes.

<<>>=
myInterestingGenes <- sample(geneNames, length(geneNames) / 10)
geneList <- factor(as.integer(geneNames %in% myInterestingGenes))
names(geneList) <- geneNames
str(geneList)
@ 

The {\tt geneList} object is a named factor that indicates which genes are interesting and which not. It
should be straightforward to compute such a named vector in a real case situation, where the user has his own
predefined list of interesting genes.

We now have all the elements to construct a {\tt topGOdata} object.



To build the {\tt topGOdata} object, we will use the MF ontology. The mapping is given by the {\tt geneID2GO}
list which will be used with the {\tt annFUN.gene2GO} function.

<<results = hide>>=
GOdata <- new("topGOdata", ontology = "MF", allGenes = geneList,
              annot = annFUN.gene2GO, gene2GO = geneID2GO)
@ 

The building of the {\tt GOdata} object can take some time, depending on the number of annotated
genes and on the chosen ontology. In our example the running time is quite fast given that we have
a rather small size gene universe which also imply a moderate size GO ontology, especially since we 
are using the MF ontology.

The advantage of having (information on) the gene scores (or better genes measurements) as well as a way to 
define which are the interesting genes, in the {\tt topGOdata} object is that one can apply
various group testing procedure, which let us test multiple hypothesis or tune with different parameters.

By typing {\tt GOdata} at the R prompt, the user can see a summary of the data.

<<>>=
GOdata
@ 

One important point to notice is that not all the genes that are provided by {\tt geneList}, the initial
gene universe, can be annotated to the GO. This can be seen by comparing the number of all available genes,
the genes present in {\tt geneList}, with the number of feasible genes. We are therefore forced at this
point to restrict the gene universe to the set of feasible genes for the rest of the analysis.

The summary on the GO graph shows the number of GO terms and the relations between them of the specified GO
ontology. This graph contains only GO terms which have at least one gene annotated to them.


\subsection{Using the genes score}  
In many cases the set of interesting genes can be computed based on a score assigned to all genes, 
for example based on the $p$-value returned by a study of differential expression. In this case, 
the {\tt topGOdata} object can store the genes score and a rule specifying the list of interesting genes.
The advantage of having both the scores and the procedure to select interesting genes encapsulated in the 
{\tt topGOdata} object is that the user can choose different types of tests statistics for the GO analysis
without modifying the input data. 

A typical example for the ALL dataset is the study where we need to discriminate between ALL cells
delivered from either B-cell or T-cell precursors.

<<results = hide>>=
y <- as.integer(sapply(eset$BT, function(x) return(substr(x, 1, 1) == 'T')))
table(y)
@ 

There are $\Sexpr{table(y)[1]}$ B-cell ALL samples and $\Sexpr{table(y)[1]}$ T-cell ALL samples in
the dataset. A two-sided $t$-test can by applied using the function {\tt getPvalues} (a wraping function for 
the {\tt mt.teststat} from the {\tt multtest} package). By default the function computes FDR (false discovery
rate) adjusted $p$-value in order to account for multiple testing. A different type of correction can be
specified using the {\tt correction} argument.

<<>>= 
geneList <- getPvalues(exprs(eset), classlabel = y, alternative = "greater")
@ 

{\tt geneList} is a named numeric vector. The gene identifiers are stored in the names attribute of the 
vector. This set of genes defines the gene universe.

Next, a function for specifying the list of interesting genes must be defined. This function needs
to select genes based on their scores (in our case the adjusted $p$-values) and must return
a logical vector specifying which gene is selected and which not. The function must have one argument,
named {\tt allScore} and must not depend on any attributes of this object. In this example we will
consider as interesting genes all genes with an adjusted $p$-value lower than $0.01$. This criteria is 
implemented in the following function:

<<results = hide>>=
topDiffGenes <- function(allScore) {
  return(allScore < 0.01)
}
x <- topDiffGenes(geneList)
sum(x) ## the number of selected genes
@ 

With all these steps done, the user can now build the {\tt topGOdata} object. For a short description 
of the arguments used by the {\tt initialize} function see Section~\ref{subsec:sigGenes}

<<results = hide>>=
GOdata <- new("topGOdata", 
              description = "GO analysis of ALL data; B-cell vs T-cell",
              ontology = "BP",
              allGenes = geneList,
              geneSel = topDiffGenes,
              annot = annFUN.db,
              nodeSize = 5,
              affyLib = affyLib)
@ 

It is often the case that many GO terms which have few annotated genes are detected to be significantly
enriched due to artifacts in the statistical test. These small sized GO terms are of less importance 
for the analysis and in many cases they can be omitted. By using the {\tt nodeSize} argument the user
can control the size of the GO terms used in the analysis. Once the genes are annotated to the each GO
term and the true path rule is applied the nodes with less than {\tt nodeSize} annotated genes are
removed from the GO hierarchy. We found that values between $5$ and $10$ for the {\tt nodeSize}
parameter yield more stable results. The default value for the {\tt nodeSize} parameter is $1$,
meaning that no pruning is performed.

Note that the only difference in the initialisation of an object of class {\tt topGOdata} to the case in
which we start with a predefined list of interesting genes is the use of the {\tt geneSel} argument. 
All further analysis depends only on the {\tt GOdata} object.


\subsection{Filtering and missing GO annotations}
\label{subsec:GOannotations}

Before going further with the enrichment analysis we analyse which of the probes available on the array
can be used in the analysis.

We want to see if the filtering performed in Section~\ref{subsec:preprocessing} removes important probes.
There are a total of $\Sexpr{nrow(ALL)}$ probes on the $\Sexpr{annotation(ALL)}$ chip. One assumes that only
the noisy probes, probes with low expression values or small variance across samples are filtered out from
the analysis. 

The number of probes have a direct effect on the multiple testing adjustment of $p$-values. Too many probes
will result in too conservative adjusted $p$-values which can bias the result of tests like Fisher's exact
test. Thus it is important to carefully analyse this step.

<<>>=
allProb <- featureNames(ALL)
groupProb <- integer(length(allProb)) + 1
groupProb[allProb %in% genes(GOdata)] <- 0
groupProb[!selProbes] <- 2
groupProb <- factor(groupProb, labels = c("Used", "Not annotated", "Filtered"))

tt <- table(groupProb)
tt
@ 



\begin{figure}[!t]
  \centering 
  \includegraphics[width=.8\linewidth]{whichProbe}
\caption{Scatter plot of FDR adjusted $p$-values against variance of probes. Points below the horizontal line
  are significant probes.}
\label{whichProbes}
\end{figure}



Out of the filtered probes only $\Sexpr{ceiling(tt[1] * 100 / sum(tt[1:2]))}\%$ have annotation to GO terms.
The filtering procedure removes $\Sexpr{tt[3]}$ probes which is a very large percentage of probes (more than 
$50\%$), but we did this intentionally to reduce the expression set for computational purposes. 

We perform a differential expression analysis on all available probes and we check if differentially 
expressed genes are leaved out from the enrichment analysis. 

<<eval = FALSE>>=
pValue <- getPvalues(exprs(ALL), classlabel = y, alternative = "greater")
geneVar <- apply(exprs(ALL), 1, var)
dd <- data.frame(x = geneVar[allProb], y = log10(pValue[allProb]), groups = groupProb)
xyplot(y ~ x | groups, data = dd, groups = groups)
@ 


<<echo = FALSE, results = hide>>=
pValue <- getPvalues(exprs(ALL), classlabel = y, alternative = "greater")
geneVar <- apply(exprs(ALL), 1, var)
dd <- data.frame(x = geneVar[allProb], y = log10(pValue[allProb]), groups = groupProb)

library(lattice)
trellis.device(device = pdf, theme = col.whitebg(), file = "whichProbe.pdf", width = 9, height = 7)
legendLab <- paste(names(table(groupProb)), " (#", table(groupProb), ")", sep = "")
pP <- xyplot(y ~ x | groups, data = dd, groups = groups,
             xlab = "Variance", ylab = "Log of p-values",
             layout = c(2, 2),
             key = list(text = list(lab = legendLab),
               points = list(pch = 20, cex = 2, 
                 col = Rows(trellis.par.get("superpose.symbol"), 1:3)$col),
               size = 7, padding.text = 3,
               x = .65, y = .7, corner = c(0, 0), border = TRUE, cex = 1),
             panel = function(x, y, ...) {
               selY <- y <= -2
               panel.xyplot(x[selY], y[selY], pch = 2, ...)
               panel.xyplot(x[!selY], y[!selY], pch = 20, ...)
               panel.abline(h = -2, lty = 2, col = "black")
             })
print(pP)
dev.off()
@ 

Figure~\ref{whichProbes} shows for the three groups of probes the adjusted $p$-values and the
gene-wise variance. Probes with large changes between conditions have large variance and low $p$-value.
In an ideal case, one would expect to have a large density of probes in the lower right corner of
{\bf Used} panel and few probes in this region in the other two panels. We can see that the filtering process
throws out some significant probes and in a real analysis a more conservative filtering needs to be applied. 
However, there are also many differentially expressed probes without GO annotation which cannot be used in
the analysis.






\section{Working with the {\tt topGOdata} object}

Once the {\tt topGOdata} object is created the user can use various methods defined for this class to
access the information encapsulated in the object.

The {\tt description} slot contains information about the experiment. This information can be accessed or
replaced using the method with the same name. 
<<results = hide>>=
description(GOdata)
description(GOdata) <- paste(description(GOdata), "Object modified on:", format(Sys.time(), "%d %b %Y"), sep = " ")
description(GOdata)
@ 

Methods to obtain the list of genes that will be used in the further analysis or methods for obtaining all 
gene scores are exemplified below.

<<>>=
a <- genes(GOdata) ## obtain the list of genes
head(a)
numGenes(GOdata)
@ 

Next we describe how to retrieve the score of a specified set of genes, e.g. a set of randomly selected genes.
If the object was constructed using a list of interesting genes, then the factor vector that was provided at
the building of the object will be returned.

<<results = hide>>=
selGenes <- sample(a, 10)
gs <- geneScore(GOdata, whichGenes = selGenes) 
print(gs)
@

If the user wants an unnamed vector or the score of all genes:
<<results = hide>>=
gs <- geneScore(GOdata, whichGenes = selGenes, use.names = FALSE)
print(gs)

gs <- geneScore(GOdata, use.names = FALSE)
str(gs)
@ 

The list of significant genes can be accessed using the method {\tt sigGenes()}.
<<results = hide>>= 
sg <- sigGenes(GOdata)
str(sg)
numSigGenes(GOdata)
@ 

Another useful method is {\tt updateGenes} which allows the user to update/change the list of genes (and
their scores) from a {\tt topGOdata} object. If one wants to update the list of genes by including only the
feasible ones, one can type:

<<results = hide>>=
.geneList <- geneScore(GOdata, use.names = TRUE)
GOdata ## more available genes
GOdata <- updateGenes(GOdata, .geneList, topDiffGenes)
GOdata ## the available genes are now the feasible genes
@ 

There are also methods available for accessing information related to GO and its structure. First, we want to
know which GO terms are available for analysis and to obtain all the genes annotated to a subset of
these GO terms.

<<>>=
graph(GOdata) ## returns the GO graph

ug <- usedGO(GOdata)
head(ug) 
@

We further select $10$ random GO terms, count the number of annotated genes and obtain their annotation.

<<results = hide>>=
sel.terms <- sample(usedGO(GOdata), 10)

num.ann.genes <- countGenesInTerm(GOdata, sel.terms) ## the number of annotated genes
num.ann.genes

ann.genes <- genesInTerm(GOdata, sel.terms) ## get the annotations
head(ann.genes)
@

When the {\tt sel.terms} argument is missing all GO terms are used. The scores for all genes, possibly 
annotated with names of the genes, can be obtained using the method {\tt scoresInTerm()}.
<<results = hide>>=
ann.score <- scoresInTerm(GOdata, sel.terms)
head(ann.score)
ann.score <- scoresInTerm(GOdata, sel.terms, use.names = TRUE)
head(ann.score)
@ 

Finally, some statistics for a set of GO terms are returned by the method {\tt termStat}. As mentioned
previously, if the {\tt sel.terms} argument is missing then the statistics for all available GO terms
are returned.
<<>>=
termStat(GOdata, sel.terms)
@ 


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%



\section{Running the enrichment tests}

In this section we explain how we can run the desired enrichment method once the {\tt topGOdata}
object is available.

{\tt topGO} package was designed to work with various test statistics and various algorithms
which take the GO dependencies into account. At the base of this design stands a S4 class
mechanism which facilitates defining and executing a (new) group test.
Three types of enrichment tests can be distinguish if we look at the data used by the each test.
\begin{itemize}
\item Tests based on gene {\bf counts}. This is the most popular family of tests, given that 
  it only requires the presence of a list of interesting genes and nothing more. Tests like 
  Fisher's exact test, Hypegeometric test and binomial test belong to this family. \cite{Draghici2006}
\item Tests based on gene {\bf scores} or gene {\bf ranks}. It includes 
  Kolmogorov-Smirnov like tests (also known as GSEA), Gentleman's Category, $t$-test, etc. \cite{Ackermann2009}
\item Tests based on gene {\bf expression}. Tests like Goeman's globaltest or GlobalAncova separates
  from the others since they work directly on the expression matrix. \cite{Goeman2007}
\end{itemize}
There are also a number of strategies/algorithms to account for the GO topology, see
Table~\ref{tabletopGO}, each of them having specific requirements.  

\begin{figure}[!b]
\centering 
\includegraphics[width=.8\linewidth]{topGO_classes_v3}
\caption{The test statistics class structure.}
\label{fig:topGOclasses}
\end{figure}

For each test type described above and for each algorithm there is S4 class defined in the
package. The main idea is to have a class (container) which can store, for a specified gene
set(GO category), all data necessary for computing the desired test statistic, and a method  
that will iterate over all GO categories. In such a design the user needs to instantiate an 
object from the class corresponding to the chosen method(test statistic and algorithm) and 
then run the iterator function on this object.

The defined S4 classes are organised in a hierarchy which is showed in Figure~\ref{fig:topGOclasses}.

% {\it The advantage of this design is that it ensures that each test statistics is used with the
%   appropriate algorithms}.

There are two possibilities(or interfaces) for applying a test statistic to an object of class
{\tt topGOdata}. The basic interface, which provides the core of the testing procedure in {\tt topGO}, 
offers more flexibility to the experienced {\tt R} user allowing him to implement new test statistics
or new algorithms. The second interface is more user friendly but at the same time more restrictive
in the choice of the tests and algorithms used. We will further explain how these two interfaces work.


\subsection{Defining and running the test}
\label{sub:define_test}

The main function for running the GO enrichment is {\tt getSigGroups()} and it takes two arguments.
The first argument is an instance of class {\tt topGOdata} and the second argument is an instance
of class {\tt groupStats} or any of its descendents.

{\it To better understand this principle consider the following example. Assume we decided to apply the
{\sf classic} algorithm. The two classes defined for this algorithm are {\tt classicCount} and
{\tt classicScore}. If an object of this class is given as a argument to {\tt getSigGroups()} than 
the classic algorithm will be used. The {\tt getSigGroups()} function can take a while, depending on the
size of the graph (the ontology used), so be patient. 
}


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection*{The {\tt groupStats} classes}
Next we show how an instance of the {\tt groupStats} class can represent a gene set and how the test
statistic is performed. 

We compute the enrichment of {\it cellular lipid metabolic process} (GO:0044255) term using 
Fisher's exact test. In order to do this we need to define the gene universe, to obtain the
genes annotated to GO:0044255, and to define the set of significant genes.

<<>>=
goID <- "GO:0044255"
gene.universe <- genes(GOdata)
go.genes <- genesInTerm(GOdata, goID)[[1]]
sig.genes <- sigGenes(GOdata)
@ 

Now we can instantiate an object of class {\tt classicCount}. Once the object is constructed we can 
get the $2\times2$ contingency table or apply the test statistic. 

<<>>=
my.group <- new("classicCount", testStatistic = GOFisherTest, name = "fisher",
                allMembers = gene.universe, groupMembers = go.genes,
                sigMembers = sig.genes)

contTable(my.group)
runTest(my.group)
@ 

The slot {\tt testStatistic} contains the function (or to be more precise, the method) which computes
the test statistic. We used the {\tt GOFisherTest} function which is available in the {\tt topGO} package
and as the name states it implements Fisher's exact test. The user can define his own test statistic
function and then apply it using the preferred algorithm. The function, however, should use the
methods defined for the {\tt groupStats} class to access the data encapsulated in such an object.
(For example a function which computes the $Z$ score can be easily implemented using as an example the
{\tt GOFisherTest} method.)

The {\tt runTest} method is defined for the {\tt groupStats} class and its used to run/compute the test
statistic, by calling the {\tt testStatistic} function. The value returned by the {\tt runTest} method
in this case is the value returned by {\tt GOFisherTest} method, which is the Fisher's exact test
$p$-value.
The {\tt contTable} method, showed in the example above, is only defined for the classes based on gene
counts and its used to compile the two-dimensional contingency table based on the object. 

To show how the same interface is used for the classes based on gene counts we next build an instance for 
the {\tt elimCount} class. We randomly select $25\%$ of the annotated genes as genes that should be removed.

<<>>=
set.seed(123)
elim.genes <- sample(go.genes, length(go.genes) / 4)
elim.group <- new("elimCount", testStatistic = GOFisherTest, name = "fisher",
                allMembers = gene.universe, groupMembers = go.genes,
                sigMembers = sig.genes, elim = elim.genes)

contTable(elim.group)
runTest(elim.group)
@ 

We see that the interface accounts for the genes that need to be eliminated, once the object is instantiated.
The same mechanism applies for the other hierarchy of classes (the score based and expression based classes), 
except that each hierarchy has its own specialised methods for computing statistics from the data.

Please note that the {\tt groupStats} class or any descendent class does not depend on GO, and an object of
such a class can be instantiated using any gene set. 


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection*{Performing the test}

According to the mechanism described above, one first defines a test statistic for the chosen algorithm, 
meaning that an instance of object specific for the algorithm is constructed in which only the test 
statistic must be specified, and then calls a {\it generic function} (interface) to run the algorithm.



{\it 
According to this mechanism, one first defines a test statistic for the chosen algorithm, in this
case {\sf classic} and then runs the algorithm (see the second line). The slot {\tt testStatistic}
contains the test statistic function. In the above example {\tt GOFisherTest} function which implements
Fisher's exact test and is available in the {\tt topGO} package was used. A user can define his own test
statistic function and then apply it using the {\sf classic} algorithm. (For example a function which
computes the $Z$ score can be implemented using as an example the {\tt GOFisherTest} method.)

}



<<results = hide>>=
test.stat <- new("classicCount", testStatistic = GOFisherTest, name = "Fisher test")
resultFisher <- getSigGroups(GOdata, test.stat)
@ 

%% {\tt getSigGroups} returns an object of class {\tt topGOresult}.
A short summary on the used test and the results is printed at the R console.

<<>>=
resultFisher
@ 
 
To use the Kolmogorov-Smirnov (KS) test one needs to provide the gene-wise scores and thus we need
to instantiate an object of a class which is able to deal of the scores. Such a class is the 
{\tt classicScore} class, see Figure~\ref{fig:topGOclasses} which will let us run the {\sf classic}
algorithm.

<<results = hide>>=
test.stat <- new("classicScore", testStatistic = GOKSTest, name = "KS tests")
resultKS <- getSigGroups(GOdata, test.stat)
@ 

%Please note that there are extra parameters for {\sf elim} and {\sf weight}, but we won't discuss them at this point. 
%This time we used the class {\tt classicScore}. This is done since the KS test needs scores of all genes and
%in this case the {\it representation} of a group of genes (GO term) is different. 

The mechanism presented above for {\sf classic} also hold for {\sf elim} and {\sf weight}. The user
should pay attention to the compatibility between the chosen class and the function for computing 
the test statistic, since no incompatibility test are made when the object is instantiated. For 
example the {\sf weight} algorithm will not work with classes based on gene-wise scores.
To run the {\sf elim} algorithm with KS test one needs to type: 

<<eval = FALSE, results = hide>>=
test.stat <- new("elimScore", testStatistic = GOKSTest, name = "Fisher test", cutOff = 0.01)
resultElim <- getSigGroups(GOdata, test.stat)
@ 

Similarly, for the {\sf weight} algorithm with Fisher's exact test one types:

<<results = hide>>=
test.stat <- new("weightCount", testStatistic = GOFisherTest, name = "Fisher test", sigRatio = "ratio")
resultWeight <- getSigGroups(GOdata, test.stat)
@ 


\subsection{The adjustment of $p$-values}

The $p$-values return by the {\tt getSigGroups} function are row $p$-values. There is no multiple testing
correction applied to them, unless the test statistic directly incorporate such a correction. Of course,
the researcher can perform an adjustment of the $p$-values if he considers it is important for the analysis.
The reason for not automatically correcting for multiple testing are:

\begin{itemize}
\item In many cases the row $p$-values return by an enrichment analysis are not that extreme. 
  %%Also, there are many terms with similar  $p$-values.
  A FDR/FWER adjustment procedure can in this case produce very conservative $p$-values and declare no,
  or very few, terms as significant. This is not necessary a bad thing, but it can happen that there are
  interesting GO terms which didn't make it over the cutoff but they are omitted and thus valuable
  information lost. In this case the researcher might be interested in the ranking of the GO terms even
  though no top term is significant at a specify FDR level.
  
\item  One should keep in mind that an enrichment analysis consist of many steps and there are many
  assumptions done before applying, for example, Fisher's  exact test on a set of GO terms. Performing
  a multiple testing procedure accounting only on the number of GO terms is far from being enough to
  control the error rate.

\item For the methods that account for the GO topology like {\sf elim} and {\sf weight}, the problem of
  multiple testing is even more complicated. Here one computes the $p$-value of a GO term conditioned on
  the neighbouring terms. The tests are therefore not independent and the multiple testing theory does
  not directly apply. We like to interpret the $p$-values returned by these methods as corrected or not
  affected by multiple testing.
\end{itemize}



\subsection{Adding a new test}
%%section to show how a test statistic can be implemented.
{\large\bf Example for the Category test ....}



\subsection{{\tt runTest}: a high-level interface for testing}
Over the basic interface we implemented an abstract layer to provide the users with a higher level
interface for running the enrichment tests. The interface is composed by a function, namely the
{\tt runTest} function, which can be used only with a predefined set of test statistics and 
algorithms. In fact {\tt runTest} is a warping function for the set of commands used for defining
and running a test presented in Section~\ref{sub:define_test}.

%%In fact the {\tt runTest} function is a warping of the {\tt getSigGroups} and the 
%%initialisation of a {\tt groupStats} object functions. 

There are three main arguments that this function takes. The first argument is an object of class
{\tt topGOdata}. The second argument, named {\tt algorithm}, is of type character and specifies which
method for dealing with the GO graph structure will be used. The third argument, named {\tt statistic},
specifies which group test statistic will be used.

To perform a classical enrichment analysis by using the {\sf classic} method and Fisher's exact test, 
the user needs to type:

<<results = hide>>=
resultFis <- runTest(GOdata, algorithm = "classic", statistic = "fisher")
@ 

Various algorithms can be easily combine with various test statistics. However not all the combinations 
will work, as seen in Table~\ref{tabletopGO}. In the case of a mismatch the function will throw an error.
The {\tt algorithm} argument is optional and if not specified the {\sf weight01} method will be used. 
Bellow we can see more examples using the {\tt runTest} function.

<<eval = FALSE, keep.source = TRUE>>=
weight01.fisher <- runTest(GOdata, statistic = "fisher")
weight01.t <- runTest(GOdata, algorithm = "weight01", statistic = "t")
elim.ks <- runTest(GOdata, algorithm = "elim", statistic = "ks")

weight.ks <- runTest(GOdata, algorithm = "weight", statistic = "ks") #will not work!!!
@ 

The last line will return an error because we cannot use the {\sf weight} method with the Kolmogorov-Smirnov
test. The methods and the statistical tests which are accessible via the {\tt runTest} function are
available via the following two functions:

<<>>=
whichTests()
whichAlgorithms()
@ 

There is no advantage of using the runTest() over getSigGroups() except that it is more user friendly and 
it gives cleaner code. However, if the user wants to define his own test statistic or implement a new 
algorithm based on the available {\tt groupStats} classes, then it would be not possible to use the
{\tt runTest} function. 

Finally, the function can pass extra arguments to the initialisation method for an {\tt groupStats} object. 
Thus, one can specify different cutoffs for the {\sf elim} method, or arguments for the {\sf weight} method. 


\section{Interpretation and visualization of results}

This section present the available tools for analysing and interpreting the results of the performed tests.
Both {\tt getSigGroups} and {\tt runTest} functions return an object of type {\tt topGOresult}, and most of
the following functions work with this object.

\subsection{The {\tt topGOresult} object}

The structure of the {\tt topGOresult} object is quite simple. It contains the $p$-values or the statistics
returned by the test and basic informations on the used test statistic/algorithm. The information stored in
the {\tt topGOdata} object is not carried over this object, and both of these objects will be needed 
by the diagnostic tools. 

{\it Since the test statistic can return either a $p$-value or a statistic of the data, we will refer them as scores!}

To access the stored $p$-values, the user should use the function {\tt score}. It returns a named numeric 
vector, were the names are GO identifiers. For example, we can look at the histogram of the results
of the Fisher's exact test and the {\sf classic} algorithm.

\setkeys{Gin}{width=.4\linewidth}
\begin{figure}[!h]
  \centering 
<<fig = TRUE>>=
pvalFis <- score(resultFis)
head(pvalFis)
hist(pvalFis, 50, xlab = "p-values")
@ 
\end{figure}

By default, the {\tt score} function does not warranty the order in which the $p$-values are returned, 
as we can see if we compare the {\tt resultFis} object with the {\tt resultWeight} object:

<<>>=
head(score(resultWeight))
@ 

However, the {\tt score} method has a parameter, {\tt whichGO}, which takes a list of GO identifiers and
returns the scores for these terms in the specified order. Only the scores for the terms found in the
intersection between the specified GOs and the GOs stored in the {\tt topGOresult} object are returned.
To see how this work lets compute the correlation between the $p$-values of the {\sf classic} and {\sf weight}
methods:

<<>>=
pvalWeight <- score(resultWeight, whichGO = names(pvalFis))
head(pvalWeight)
cor(pvalFis, pvalWeight)
@ 

Basic information on input data can be accessed using the {\tt geneData} function. The number of annotated genes, 
the number of significant genes (if it is the case), the minimal size of a GO category as well as the number 
of GO categories which have at least one significant gene annotated are listed:

<<>>=
geneData(resultWeight)
@ 


\subsection{Summarising the results}

We can use the {\tt GenTable} function to generate a summary table with the results from one or more tests
applied to the same {\tt topGOdata} object. The function can take a variable number of {\tt topGOresult} objects
and it returns a {\tt data.frame} containing the top {\tt topNodes} GO terms identified by the method specified 
with the {\tt orderBy} argument. This argument allows the user decide which $p$-values should be used for
ordering the GO terms. 

<<>>=
allRes <- GenTable(GOdata, classic = resultFis, KS = resultKS, weight = resultWeight,
                   orderBy = "weight", ranksOf = "classic", topNodes = 20)
@ 

Please note that we need to type the full names (the exact name) of the function arguments: {\tt topNodes},
{\tt rankOf}, etc. This is the price paid for flexibility of specifying different number of {\tt topGOresults} 
objects. The table includes statistics on the GO terms plus the $p$-values returned by the other algorithms/test
statistics. Table~\ref{tab:GOresults} shows the informations included in the {\tt data.frame}.

\begin{table}[!t]
  \centering\resizebox{.99\linewidth}{!}{%
<<echo = FALSE, results = tex>>=
if(require(xtable))
  print(xtable(apply(allRes, 2, as.character)), floating = FALSE)
@
}\caption{Significance of GO terms according to different tests.}
\label{tab:GOresults}
\end{table}


\subsection{Analysing individual GOs}
Next we want to analyse the distribution of the genes annotated to a GO term of interest. In 
an enrichment analysis one expects that the genes annotated to a significantly enriched GO term
have higher scores than the average gene' score of the gene universe. 

One way to check this hypothesis is to compare the distribution of the gene scores annotated to the 
specified GO term with the distribution of the scores of the complementary gene set (all the genes 
in the gene universe which are not annotated to the GO term). This can be easily achieved using the 
{\tt showGroupDensity} function. For example, lets look at the distribution of the genes annotated to 
the most significant GO term w.r.t. the {\sf weight} algorithm. 

\setkeys{Gin}{width=.4\linewidth}
\begin{figure}[!bh]
  \centering 
<<fig = TRUE>>=
goID <- allRes[1, "GO.ID"]
print(showGroupDensity(GOdata, goID, ranks = TRUE))
@ 
\caption{Distribution of the gene' rank from \Sexpr{goID}, compared with the null distribution.}
\label{fig:geneDensityDiff}
\end{figure}

We can see in Figure~\ref{fig:geneDensityDiff} that the genes annotated to \Sexpr{goID} have low ranks
(genes with low $p$-value of the $t$-test). The distribution of the ranks is skewed on the left side 
compared with the reference distribution given by the complementary gene set. This is a nice example
in which there is a significant difference in the distribution of scores between the gene set and the 
complementary set, and we see from Table~\ref{tab:GOresults} that this GO is found as significantly 
enriched by all methods used.

In the above example, the genes with a $p$-value equal to $1$ were omitted. They can be included using 
the value {\tt FALSE} for the {\tt rm.one} argument in the {\tt showGroupDensity} function.


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% the printGenes function
Another useful function for analysing terms of interest is {\tt printGenes}. The function will generate 
a table with all the genes/probes annotated to the specified GO term. Various type of identifiers, the 
gene name and the $p$-values/statistics are provided in the table. 

<<>>=
goID <- allRes[10, "GO.ID"]
gt <- printGenes(GOdata, whichTerms = goID, chip = affyLib, numChar = 40)
@ 

\begin{table}[!ht]
  \centering\resizebox{.99\linewidth}{!}{%
<<echo = FALSE, results = tex>>=
if(require(xtable))
  print(xtable(gt), floating = FALSE)
@
}\caption{Genes annotated to \Sexpr{goID}.}
\label{tab:printGenes1}
\end{table}

The {\tt data.frame} containing the genes annotated to \Sexpr{goID} is shown in Table~\ref{tab:printGenes1}.
One or more GO identifiers can be given to the function using the {\tt whichTerms} argument. When more than
one GO is specified, the function returns a list of data.frames, otherwise only one {\tt data.frame} is returned.
{\it The function has a argument {\sf file} which, when specified, will save the results into a file using the 
CSV format.}

For the moment the function will work only when the chip used has an annotation package available in
Bioconductor. It will not work with other type of custom annotations. 



\subsection{Visualising the GO structure}
An insightful way of looking at the results of the analysis is to investigate how the significant GO
terms are distributed over the GO graph. We plot the subgraphs induced by the most significant GO terms 
reported by {\sf classic} and {\sf weight} methods. There are two functions available. The {\tt showSigOfNodes}
will plot the induced subgraph to the current graphic device. The {\tt printGraph} is a warping function
of {\tt showSigOfNodes} and will save the resulting graph into a PDF or PS file. 

<<eval = FALSE>>=
showSigOfNodes(GOdata, score(resultFis), firstSigNodes = 5, useInfo = 'all')
showSigOfNodes(GOdata, score(resultWeight), firstSigNodes = 5, useInfo = 'def')
@ 

<<results = hide>>=
printGraph(GOdata, resultFis, firstSigNodes = 5, fn.prefix = "tGO", useInfo = "all", pdfSW = TRUE)
printGraph(GOdata, resultWeight, firstSigNodes = 5, fn.prefix = "tGO", useInfo = "def", pdfSW = TRUE)
@ 

In the plots, the {\em significant nodes} are represented as rectangles. The plotted graph is
the upper induced graph generated by these {\em significant nodes}. 
These graph plots are used to see how the significant GO terms are distributed in the hierarchy. It is a
very useful tool to realize behaviour of various enrichment methods and to better understand which of the
significant GO terms are really of interest. 

\begin{figure}[!t]
\centering 
\includegraphics[width=1.05\linewidth]{tGO_classic_5_all}
\caption{The subgraph induced by the top 5 GO terms identified by the {\sf classic} algorithm for
  scoring GO terms for enrichment. Boxes indicate the 5 most significant terms. Box color represents the
  relative significance, ranging from dark red (most significant) to light yellow (least significant).
  Black arrows indicate {\it is-a} relationships and red arrows {\it part-of} relationships.}
\label{fig:GOclassic}
\end{figure}


\begin{figure}[!t]
\label{fig:GOweight}\centering 
\includegraphics[width=1.05\linewidth]{tGO_weight_5_def}
\caption{The subgraph induced by the top 5 GO terms identified by the {\sf weight} algorithm for
  scoring GO terms for enrichment. Boxes indicate the 5 most significant terms. Box color represents the
  relative significance, ranging from dark red (most significant) to light yellow (least significant).
  Black arrows indicate {\it is-a} relationships and red arrows {\it part-of} relationships.}
\end{figure}

We can emphasise differences between two methods using the {\tt printGraph} function:
<<eval = FALSE>>=
printGraph(GOdata, resultWeight, firstSigNodes = 10, resultFis, fn.prefix = "tGO", useInfo = "def")
printGraph(GOdata, resultElim, firstSigNodes = 15, resultFis, fn.prefix = "tGO", useInfo = "all")
@ 

\clearpage

\section{Session Information}

The version number of R and packages loaded for generating the vignette were:

<<echo=FALSE,results=tex>>=
toLatex(sessionInfo())
@


\addcontentsline{toc}{section}{References} \label{references}


\bibliography{ref_books}
\bibliographystyle{apalike}
\nocite*

\end{document}