Package elki.clustering.biclustering

Class ChengAndChurch

java.lang.Object

elki.clustering.biclustering.AbstractBiclustering<BiclusterWithInversionsModel>

elki.clustering.biclustering.ChengAndChurch

All Implemented Interfaces:

Algorithm, ClusteringAlgorithm<Clustering<BiclusterWithInversionsModel>>

@Reference(authors="Y. Cheng, G. M. Church",
           title="Biclustering of expression data",
           booktitle="Proc. 8th Int. Conf. on Intelligent Systems for Molecular Biology (ISMB)",
           url="http://www.aaai.org/Library/ISMB/2000/ismb00-010.php",
           bibkey="DBLP:conf/ismb/ChengC00")
public class ChengAndChurch
extends AbstractBiclustering<BiclusterWithInversionsModel>

Cheng and Church biclustering.

Reference:

Y. Cheng and G. M. Church.
Biclustering of expression data.
Proc. 8th Int. Conf. on Intelligent Systems for Molecular Biology (ISMB)

Since:

0.6.0

Author:

Erich Schubert

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
protected static class	ChengAndChurch.BiclusterCandidate	Bicluster candidate.
protected static interface	ChengAndChurch.CellVisitor	Visitor pattern for processing cells.

Nested classes/interfaces inherited from interface elki.Algorithm

Algorithm.Utils

Field Summary

Fields

Modifier and Type	Field	Description
private double	alpha	The parameter for multiple node deletion.
private double	delta	Threshold for the score.
private Distribution	dist	Distribution to sample random replacement values from.
private static Logging	LOG	The logger for this class.
private static int	MIN_COLUMN_REMOVE_THRESHOLD	The minimum number of columns that the database must have so that a removal of columns is performed in multipleNodeDeletion(double[][], elki.clustering.biclustering.ChengAndChurch.BiclusterCandidate).
private static int	MIN_ROW_REMOVE_THRESHOLD	The minimum number of rows that the database must have so that a removal of rows is performed in multipleNodeDeletion(double[][], elki.clustering.biclustering.ChengAndChurch.BiclusterCandidate).
private int	n	Number of biclusters to be found.
private RandomFactory	rnd	Random generator
private boolean	useinverted	Allow inversion of rows in the last phase.

Fields inherited from class elki.clustering.biclustering.AbstractBiclustering

relation, rowIDs

Constructor Summary

Constructors

Constructor	Description
ChengAndChurch(double delta, double alpha, int n, Distribution dist, RandomFactory rnd)	Constructor.

Method Summary

Modifier and Type	Method	Description
Clustering<BiclusterWithInversionsModel>	biclustering()	Run the actual biclustering algorithm.
TypeInformation[]	getInputTypeRestriction()	Get the input type restriction used for negotiating the data query.
private void	multipleNodeDeletion(double[][] mat, ChengAndChurch.BiclusterCandidate cand)	Algorithm 2 of Cheng and Church.
private void	nodeAddition(double[][] mat, ChengAndChurch.BiclusterCandidate cand)	Algorithm 3 of Cheng and Church.
private void	singleNodeDeletion(double[][] mat, ChengAndChurch.BiclusterCandidate cand)	Algorithm 1 of Cheng and Church:

Methods inherited from class elki.clustering.biclustering.AbstractBiclustering

colsBitsetToIDs, colsBitsetToIDs, defineBicluster, defineBicluster, getColDim, getRowDBID, getRowDim, rowsBitsetToIDs, rowsBitsetToIDs, run, valueAt

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface elki.clustering.ClusteringAlgorithm

autorun

Field Detail

LOG

private static final Logging LOG

The logger for this class.

MIN_COLUMN_REMOVE_THRESHOLD

private static final int MIN_COLUMN_REMOVE_THRESHOLD

The minimum number of columns that the database must have so that a removal of columns is performed in multipleNodeDeletion(double[][], elki.clustering.biclustering.ChengAndChurch.BiclusterCandidate).

Just start deleting multiple columns when more than 100 columns are in the data matrix.

See Also:

Constant Field Values

MIN_ROW_REMOVE_THRESHOLD

private static final int MIN_ROW_REMOVE_THRESHOLD

The minimum number of rows that the database must have so that a removal of rows is performed in multipleNodeDeletion(double[][], elki.clustering.biclustering.ChengAndChurch.BiclusterCandidate).

Just start deleting multiple rows when more than 100 rows are in the data matrix.

See Also:

Constant Field Values

delta

private double delta

Threshold for the score.

alpha

private double alpha

The parameter for multiple node deletion.

It is used to magnify the delta value in the multipleNodeDeletion(double[][], elki.clustering.biclustering.ChengAndChurch.BiclusterCandidate) method.

n

private int n

Number of biclusters to be found.

useinverted

private boolean useinverted

Allow inversion of rows in the last phase.

dist

private Distribution dist

Distribution to sample random replacement values from.

rnd

private RandomFactory rnd

Random generator

Constructor Detail

ChengAndChurch

public ChengAndChurch(double delta,
                      double alpha,
                      int n,
                      Distribution dist,
                      RandomFactory rnd)

Constructor.

Parameters:

delta - Delta parameter: desired quality

alpha - Alpha parameter: controls switching to single node deletion approach

n - Number of clusters to detect

dist - Distribution of random values to insert

rnd - Random factory

Method Detail

biclustering

public Clustering<BiclusterWithInversionsModel> biclustering()

Description copied from class: AbstractBiclustering

Run the actual biclustering algorithm.

This method is supposed to be called only from the method AbstractBiclustering.run(elki.database.relation.Relation<? extends elki.data.NumberVector>).

Specified by:

biclustering in class AbstractBiclustering<BiclusterWithInversionsModel>

singleNodeDeletion

private void singleNodeDeletion(double[][] mat,
                                ChengAndChurch.BiclusterCandidate cand)

Algorithm 1 of Cheng and Church:

Remove single rows or columns.

Inverted rows are not supported in this method.

Parameters:

mat - Data matrix

cand - Bicluster candidate

multipleNodeDeletion

private void multipleNodeDeletion(double[][] mat,
                                  ChengAndChurch.BiclusterCandidate cand)

Algorithm 2 of Cheng and Church.

Remove all rows and columns that reduce the residue by alpha.

Inverted rows are not supported in this method.

Parameters:

mat - Data matrix

cand - Bicluster candidate

nodeAddition

private void nodeAddition(double[][] mat,
                          ChengAndChurch.BiclusterCandidate cand)

Algorithm 3 of Cheng and Church.

Try to re-add rows or columns that decrease the overall score.

Also try adding inverted rows.

Parameters:

mat - Data matrix

cand - Bicluster candidate

getInputTypeRestriction

public TypeInformation[] getInputTypeRestriction()

Description copied from interface: Algorithm

Get the input type restriction used for negotiating the data query.

Returns:

Type restriction