GroupHypAlgorithm Class Reference

This class implements an enrichment method similar to the GenGO method (see http://www.ncbi.nlm.nih.gov/pubmed/18676451). However, instead of maximizing the log-likelihood function discussed in the GenGO paper, GroupHypAlgorithm minimizes the hypergeometric p-value for a given set of functions. Initially, all functions are inactive. We try all possible one-function changes (i.e., remove or add a single function) to the current set of active functions. By treating the collection of active functions as a single function, we can see if the addition or removal of any function causes the group p-value to decrease. If so, we add or remove that function and repeat. More...

#include <enrichment-algorithm.h>

Inheritance diagram for GroupHypAlgorithm:

List of all members.

Public Member Functions
	GroupHypAlgorithm (const MyGraph &clusterGraph, const MyGraph &universeGraph, MyAnnotations &annotations, const GeneOntology &go, string category)
	Default constructor.
virtual	~GroupHypAlgorithm ()
	Default destructor.
virtual void	computeEnrichedFunctions (const set< string > &functionsToProcess)
	Computes the enrichment of all the functions.
virtual void	computeEnrichmentForFunction (const string funcType, const string funcId)
	Computes whether one function is enriched.
virtual void	printEnrichmentResults (ostream &ostr)
	prints the enrichment results

---

Detailed Description

This class implements an enrichment method similar to the GenGO method (see http://www.ncbi.nlm.nih.gov/pubmed/18676451). However, instead of maximizing the log-likelihood function discussed in the GenGO paper, GroupHypAlgorithm minimizes the hypergeometric p-value for a given set of functions. Initially, all functions are inactive. We try all possible one-function changes (i.e., remove or add a single function) to the current set of active functions. By treating the collection of active functions as a single function, we can see if the addition or removal of any function causes the group p-value to decrease. If so, we add or remove that function and repeat.

---

Constructor & Destructor Documentation

GroupHypAlgorithm::GroupHypAlgorithm	(	const MyGraph &	clusterGraph,
		const MyGraph &	universeGraph,
		MyAnnotations &	annotations,
		const GeneOntology &	go,
		string	category
	)

Default constructor.

Parameters:

[in]	clusterGraph	an instance of MyGraph storing the "cluster" for which we want to compute functional enrichment.
[in]	universeGraph	an instance of MyGraph storing the universe of genes.
[in]	annotations	The annotations for the genes.

Warning:

clusterGraph should be a subgraph of universeGraph, but the class does not check this condition.

---

Member Function Documentation

void GroupHypAlgorithm::computeEnrichedFunctions

(

const set< string > &

functionsToProcess

)

[virtual]

Computes the enrichment of all the functions.

Parameters:

[in]

functionstoProcess,a

set of strings, where each string represents a function ID. If the set is nonempty, functional enrichment will only be performed for the listed functions in this set.

Reimplemented from FunctionalEnrichmentAlgorithm.

virtual void GroupHypAlgorithm::computeEnrichmentForFunction	(	const string	funcType,
		const string	funcId
	)		[inline, virtual]

Computes whether one function is enriched.

Parameters:

[in]	type	a string specifying the type of the function.
[in]	functionId	a string specifying function.

Implements FunctionalEnrichmentAlgorithm.

void GroupHypAlgorithm::printEnrichmentResults

(

ostream &

ostr

)

[virtual]

prints the enrichment results

Parameters:

[in]

ostr,an

output stream to which the results will be written. Use cout for printing to a terminal.

Implements FunctionalEnrichmentAlgorithm.

---

The documentation for this class was generated from the following files:

• /home/poirel/src/c++/biorithm/libfun/enrichment-algorithm.h
• /home/poirel/src/c++/biorithm/libfun/enrichment-algorithm.C