Hierarchical clustering. More...

#include <Bpp/Phyl/Distance/HierarchicalClustering.h>

Inheritance diagram for bpp::HierarchicalClustering:

Collaboration diagram for bpp::HierarchicalClustering:

Public Member Functions
	HierarchicalClustering (const std::string &method, bool verbose=false)
	Builds a new clustering object. More...

	HierarchicalClustering (const std::string &method, const DistanceMatrix &matrix, bool verbose=false)

virtual	~HierarchicalClustering ()

HierarchicalClustering *	clone () const

std::string	getName () const

virtual void	setDistanceMatrix (const DistanceMatrix &matrix) override
	Set the distance matrix to use. More...

bool	hasTree () const override

const Tree &	tree () const override

void	computeTree () override
	Compute the tree corresponding to the distance matrix. More...

void	setVerbose (bool yn) override

bool	isVerbose () const override

Static Public Attributes
static const std::string	COMPLETE = "Complete"

static const std::string	SINGLE = "Single"

static const std::string	AVERAGE = "Average"

static const std::string	MEDIAN = "Median"

static const std::string	WARD = "Ward"

static const std::string	CENTROID = "Centroid"

Protected Member Functions
std::vector< size_t >	getBestPair ()
	Get the best pair of nodes to agglomerate. More...

std::vector< double >	computeBranchLengthsForPair (const std::vector< size_t > &pair)
	Compute the branch lengths for two nodes to agglomerate. More...

double	computeDistancesFromPair (const std::vector< size_t > &pair, const std::vector< double > &branchLengths, size_t pos)
	Actualizes the distance matrix according to a given pair and the corresponding branch lengths. More...

void	finalStep (int idRoot)
	Method called when there ar eonly three remaining node to agglomerate, and creates the root node of the tree. More...

virtual Node *	getLeafNode (int id, const std::string &name)
	Get a leaf node. More...

virtual Node *	getParentNode (int id, Node son1, Node son2)
	Get an inner node. More...

Protected Attributes
std::string	method_

DistanceMatrix	matrix_

std::unique_ptr< Tree >	tree_

std::map< size_t, Node * >	currentNodes_

bool	verbose_

bool	rootTree_

Detailed Description

Hierarchical clustering.

This class implements the complete, single, average (= UPGMA), median, ward and centroid linkage methods.

Definition at line 29 of file HierarchicalClustering.h.

Constructor & Destructor Documentation

◆ HierarchicalClustering() [1/2]

bpp::HierarchicalClustering::HierarchicalClustering	(	const std::string &	method,
		bool	verbose = `false`
	)

inline

Builds a new clustering object.

Parameters

method	The linkage method to use. should be one of COMPLETE, SINGLE, AVERAGE, MEDIAN, WARD, CENTROID.
verbose	Tell if some progress information should be displayed.

Definition at line 50 of file HierarchicalClustering.h.

Referenced by clone().

◆ HierarchicalClustering() [2/2]

bpp::HierarchicalClustering::HierarchicalClustering	(	const std::string &	method,
		const DistanceMatrix &	matrix,
		bool	verbose = `false`
	)

inline

Definition at line 53 of file HierarchicalClustering.h.

References bpp::AbstractAgglomerativeDistanceMethod::computeTree().

◆ ~HierarchicalClustering()

virtual bpp::HierarchicalClustering::~HierarchicalClustering ( )

inlinevirtual

Definition at line 60 of file HierarchicalClustering.h.

Member Function Documentation

◆ clone()

HierarchicalClustering* bpp::HierarchicalClustering::clone ( ) const

inlinevirtual

Implements bpp::AgglomerativeDistanceMethodInterface.

Definition at line 62 of file HierarchicalClustering.h.

References HierarchicalClustering().

◆ computeBranchLengthsForPair()

vector< double > HierarchicalClustering::computeBranchLengthsForPair ( const std::vector< size_t > & pair )

protectedvirtual

Compute the branch lengths for two nodes to agglomerate.

+---l1-----N1
|
+---l2-----N2

This method compute l1 and l2 given N1 and N2.

Parameters

pair	The indices of the nodes to be agglomerated.

Returns: A size 2 vector with branch lengths.

Implements bpp::AbstractAgglomerativeDistanceMethod.

Definition at line 62 of file HierarchicalClustering.cpp.

◆ computeDistancesFromPair()

double HierarchicalClustering::computeDistancesFromPair	(	const std::vector< size_t > &	pair,
		const std::vector< double > &	branchLengths,
		size_t	pos
	)

protectedvirtual

Actualizes the distance matrix according to a given pair and the corresponding branch lengths.

Parameters

pair	The indices of the nodes to be agglomerated.
branchLengths	The corresponding branch lengths.
pos	The index of the node whose distance ust be updated.

Returns: The distance between the 'pos' node and the agglomerated pair.

Implements bpp::AbstractAgglomerativeDistanceMethod.

Definition at line 71 of file HierarchicalClustering.cpp.

References bpp::abs(), and bpp::pow().

◆ computeTree()

void AbstractAgglomerativeDistanceMethod::computeTree ( )

overridevirtualinherited

Compute the tree corresponding to the distance matrix.

This method implements the following algorithm: 1) Build all leaf nodes (getLeafNode method) 2) Get the best pair to agglomerate (getBestPair method) 3) Compute the branch lengths for this pair (computeBranchLengthsForPair method) 4) Build the parent node of the pair (getParentNode method) 5) For each remaining node, update distances from the pair (computeDistancesFromPair method) 6) Return to step 2 while there are more than 3 remaining nodes. 7) Perform the final step, and send a rooted or unrooted tree.

Implements bpp::DistanceMethodInterface.

Reimplemented in bpp::BioNJ.

Definition at line 26 of file AbstractAgglomerativeDistanceMethod.cpp.

References bpp::ApplicationTools::displayGauge(), and bpp::Node::setDistanceToFather().

Referenced by HierarchicalClustering(), bpp::NeighborJoining::NeighborJoining(), and bpp::PGMA::PGMA().

◆ finalStep()

void HierarchicalClustering::finalStep ( int idRoot )

protectedvirtual

Method called when there ar eonly three remaining node to agglomerate, and creates the root node of the tree.

Parameters

idRoot The id of the root node.

Implements bpp::AbstractAgglomerativeDistanceMethod.

Definition at line 131 of file HierarchicalClustering.cpp.

References bpp::Node::addSon(), and bpp::Node::setDistanceToFather().

◆ getBestPair()

vector< size_t > HierarchicalClustering::getBestPair ( )

protectedvirtual

Get the best pair of nodes to agglomerate.

Define the criterion to chose the next pair of nodes to agglomerate. This criterion uses the matrix_ distance matrix.

Returns: A size 2 vector with the indices of the nodes.

Exceptions

Exception If an error occurred.

Implements bpp::AbstractAgglomerativeDistanceMethod.

Definition at line 18 of file HierarchicalClustering.cpp.

References bpp::numeric::log().

◆ getLeafNode()

Node * HierarchicalClustering::getLeafNode	(	int	id,
		const std::string &	name
	)

protectedvirtual

Get a leaf node.

Create a new node with the given id and name.

Parameters

id	The id of the node.
name	The name of the node.

Returns: A pointer toward a new node object.

Reimplemented from bpp::AbstractAgglomerativeDistanceMethod.

Definition at line 148 of file HierarchicalClustering.cpp.

References bpp::ClusterInfos::length, bpp::ClusterInfos::numberOfLeaves, and bpp::NodeTemplate< NodeInfos >::setInfos().

◆ getName()

std::string bpp::HierarchicalClustering::getName ( ) const

inlinevirtual

Returns: The name of the distance method.

Implements bpp::DistanceMethodInterface.

Definition at line 65 of file HierarchicalClustering.h.

References method_.

◆ getParentNode()

Node * HierarchicalClustering::getParentNode	(	int	id,
		Node *	son1,
		Node *	son2
	)

protectedvirtual

Get an inner node.

Create a new node with the given id, and set its sons.

Parameters

id	The id of the node.
son1	The first son of the node.
son2	The second son of the node.

Returns: A pointer toward a new node object.

Reimplemented from bpp::AbstractAgglomerativeDistanceMethod.

Definition at line 158 of file HierarchicalClustering.cpp.

References bpp::Node::addSon(), bpp::Node::getDistanceToFather(), bpp::ClusterInfos::length, and bpp::ClusterInfos::numberOfLeaves.

◆ hasTree()

bool bpp::AbstractAgglomerativeDistanceMethod::hasTree ( ) const

inlineoverridevirtualinherited

Returns: True if a tree has been computed.

Implements bpp::DistanceMethodInterface.

Definition at line 86 of file AbstractAgglomerativeDistanceMethod.h.

References bpp::AbstractAgglomerativeDistanceMethod::tree_.

◆ isVerbose()

bool bpp::AbstractAgglomerativeDistanceMethod::isVerbose ( ) const

inlineoverridevirtualinherited

Returns: True if verbose mode is enabled.

Implements bpp::DistanceMethodInterface.

Definition at line 114 of file AbstractAgglomerativeDistanceMethod.h.

References bpp::AbstractAgglomerativeDistanceMethod::verbose_.

◆ setDistanceMatrix()

void AbstractAgglomerativeDistanceMethod::setDistanceMatrix ( const DistanceMatrix & matrix )

overridevirtualinherited

Set the distance matrix to use.

Parameters

matrix The matrix to use.

Exceptions

Exception In case an incorrect matrix is provided (eg smaller than 3).

Implements bpp::DistanceMethodInterface.

Reimplemented in bpp::PGMA, bpp::NeighborJoining, and bpp::BioNJ.

Definition at line 17 of file AbstractAgglomerativeDistanceMethod.cpp.

References bpp::DistanceMatrix::reset(), and bpp::DistanceMatrix::size().

Referenced by bpp::AbstractAgglomerativeDistanceMethod::AbstractAgglomerativeDistanceMethod(), bpp::NeighborJoining::setDistanceMatrix(), and bpp::PGMA::setDistanceMatrix().

◆ setVerbose()

void bpp::AbstractAgglomerativeDistanceMethod::setVerbose ( bool yn )

inlineoverridevirtualinherited

Parameters

yn	Enable/Disable verbose mode.

Implements bpp::DistanceMethodInterface.

Definition at line 113 of file AbstractAgglomerativeDistanceMethod.h.

References bpp::AbstractAgglomerativeDistanceMethod::verbose_.

◆ tree()

const Tree& bpp::AbstractAgglomerativeDistanceMethod::tree ( ) const

inlineoverridevirtualinherited

Returns: A reference toward the computed tree. Throws an exception if no tree was computed.

Implements bpp::DistanceMethodInterface.

Definition at line 91 of file AbstractAgglomerativeDistanceMethod.h.

References bpp::AbstractAgglomerativeDistanceMethod::tree_.

Member Data Documentation

◆ AVERAGE

const string HierarchicalClustering::AVERAGE = "Average"

static

Definition at line 35 of file HierarchicalClustering.h.

◆ CENTROID

const string HierarchicalClustering::CENTROID = "Centroid"

static

Definition at line 38 of file HierarchicalClustering.h.

◆ COMPLETE

const string HierarchicalClustering::COMPLETE = "Complete"

static

Definition at line 33 of file HierarchicalClustering.h.

◆ currentNodes_

std::map<size_t, Node*> bpp::AbstractAgglomerativeDistanceMethod::currentNodes_

protectedinherited

Definition at line 33 of file AbstractAgglomerativeDistanceMethod.h.

Referenced by bpp::AbstractAgglomerativeDistanceMethod::operator=().

◆ matrix_

DistanceMatrix bpp::AbstractAgglomerativeDistanceMethod::matrix_

protectedinherited

Definition at line 30 of file AbstractAgglomerativeDistanceMethod.h.

Referenced by bpp::AbstractAgglomerativeDistanceMethod::operator=().

◆ MEDIAN

const string HierarchicalClustering::MEDIAN = "Median"

static

Definition at line 36 of file HierarchicalClustering.h.

◆ method_

std::string bpp::HierarchicalClustering::method_

protected

Definition at line 41 of file HierarchicalClustering.h.

Referenced by getName().

◆ rootTree_

bool bpp::AbstractAgglomerativeDistanceMethod::rootTree_

protectedinherited

Definition at line 35 of file AbstractAgglomerativeDistanceMethod.h.

Referenced by bpp::AbstractAgglomerativeDistanceMethod::operator=().

◆ SINGLE

const string HierarchicalClustering::SINGLE = "Single"

static

Definition at line 34 of file HierarchicalClustering.h.

◆ tree_

std::unique_ptr<Tree> bpp::AbstractAgglomerativeDistanceMethod::tree_

protectedinherited

Definition at line 31 of file AbstractAgglomerativeDistanceMethod.h.

Referenced by bpp::AbstractAgglomerativeDistanceMethod::AbstractAgglomerativeDistanceMethod(), bpp::AbstractAgglomerativeDistanceMethod::hasTree(), bpp::AbstractAgglomerativeDistanceMethod::operator=(), and bpp::AbstractAgglomerativeDistanceMethod::tree().

◆ verbose_

bool bpp::AbstractAgglomerativeDistanceMethod::verbose_

protectedinherited

Definition at line 34 of file AbstractAgglomerativeDistanceMethod.h.

Referenced by bpp::AbstractAgglomerativeDistanceMethod::isVerbose(), bpp::AbstractAgglomerativeDistanceMethod::operator=(), and bpp::AbstractAgglomerativeDistanceMethod::setVerbose().

◆ WARD

const string HierarchicalClustering::WARD = "Ward"

static

Definition at line 37 of file HierarchicalClustering.h.

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

Bpp/Phyl/Distance/HierarchicalClustering.h
Bpp/Phyl/Distance/HierarchicalClustering.cpp

Public Member Functions

Static Public Attributes

Protected Member Functions

Protected Attributes

Detailed Description

Constructor & Destructor Documentation

◆ HierarchicalClustering() [1/2]

◆ HierarchicalClustering() [2/2]

◆ ~HierarchicalClustering()

Member Function Documentation

◆ clone()

◆ computeBranchLengthsForPair()

◆ computeDistancesFromPair()

◆ computeTree()

◆ finalStep()

◆ getBestPair()

◆ getLeafNode()

◆ getName()

◆ getParentNode()

◆ hasTree()

◆ isVerbose()

◆ setDistanceMatrix()

◆ setVerbose()

◆ tree()

Member Data Documentation

◆ AVERAGE

◆ CENTROID

◆ COMPLETE

◆ currentNodes_

◆ matrix_

◆ MEDIAN

◆ method_

◆ rootTree_

◆ SINGLE

◆ tree_

◆ verbose_

◆ WARD