de.jstacs.clustering.distances

Class SequenceScoreDistance

java.lang.Object

de.jstacs.clustering.distances.DistanceMetric<StatisticalModel>

de.jstacs.clustering.distances.SequenceScoreDistance

Direct Known Subclasses:

RandomSequenceScoreDistance

public class SequenceScoreDistance
extends DistanceMetric<StatisticalModel>

Class for a distance metric between StatisticalModels based on the correlation of score profiles on De Bruijn sequences. For two StatisticalModels and , we compute the score profiles and on a De Bruijn sequence of length . The distance is then defined based on the Pearson correlation as between these score profiles, maximizing over suitable shifts of the score profiles and both strand orientations.

Author:

Jan Grau

Field Summary

Fields

Modifier and Type	Field and Description
protected boolean	exp if exponential scores should be used
protected CyclicSequenceAdaptor[]	seqs The De Bruijn sequences

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	SequenceScoreDistance(CyclicSequenceAdaptor[] seqs, boolean exp) Creates a new distance for a given set of sequences.
	SequenceScoreDistance(DiscreteAlphabet alphabet, int n, boolean exp) Creates a new distance.

Method Summary

Methods

Modifier and Type	Method and Description
double	getDistance(double[][] profiles1, double[][] profiles1Rc, double[][] profiles2, int maxShift) Returns the distance between the two score profiles.
double	getDistance(double[][] profiles1, double[][] profiles1Rc, StatisticalModel o2, int motif1Length) Returns the distance between a score profile and a model.
double	getDistance(StatisticalModel o1, StatisticalModel o2) Returns the distance according to the metric of the two supplied objects.
double[][]	getPairwiseDistanceMatrix(int numThreads, StatisticalModel... objects) Multi-threaded computation of the pairwise distance matrix.
double[][]	getProfile(StatisticalModel o, boolean rc) Returns the score profile for the model.

Methods inherited from class de.jstacs.clustering.distances.DistanceMetric

getPairwiseDistanceMatrix

Methods inherited from class java.lang.Object

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

Field Detail

seqs

protected CyclicSequenceAdaptor[] seqs

The De Bruijn sequences

exp

protected boolean exp

if exponential scores should be used

Constructor Detail

SequenceScoreDistance

public SequenceScoreDistance(DiscreteAlphabet alphabet,
                     int n,
                     boolean exp)
                      throws WrongAlphabetException,
                             WrongSequenceTypeException

Creates a new distance.

Parameters:

alphabet - the alphabet of the models that may be compared

n - the length of n-mers represented in the De Bruijn sequence

exp - if exponential scores should be used

Throws:

WrongAlphabetException

WrongSequenceTypeException

SequenceScoreDistance

protected SequenceScoreDistance(CyclicSequenceAdaptor[] seqs,
                     boolean exp)

Creates a new distance for a given set of sequences.

Parameters:

seqs - the sequences

exp - if exponential scores should be used

Method Detail

getProfile

public double[][] getProfile(StatisticalModel o,
                    boolean rc)
                      throws Exception

Returns the score profile for the model.

Parameters:

o - the mode

rc - if the reverse complement should be considered

Returns:

the score profile

Throws:

Exception - if the score could not be computed

getDistance

public double getDistance(double[][] profiles1,
                 double[][] profiles1Rc,
                 double[][] profiles2,
                 int maxShift)
                   throws Exception

Returns the distance between the two score profiles.

Parameters:

profiles1 - the first profile

profiles1Rc - the reverse complementary version of the first profile

profiles2 - the second profile

maxShift - the maximum allowed shift between the profiles

Returns:

the distance

Throws:

Exception - if the distance could not be computed

getDistance

public double getDistance(double[][] profiles1,
                 double[][] profiles1Rc,
                 StatisticalModel o2,
                 int motif1Length)
                   throws Exception

Returns the distance between a score profile and a model.

Parameters:

profiles1 - the first profile

profiles1Rc - the reverse complementary version of the first profile

o2 - the model

motif1Length - the length of the motif used to compute the first profile

Returns:

the distance

Throws:

Exception - if the distance could not be computed

getDistance

public double getDistance(StatisticalModel o1,
                 StatisticalModel o2)
                   throws Exception

Description copied from class: DistanceMetric

Returns the distance according to the metric of the two supplied objects.

Specified by:

getDistance in class DistanceMetric<StatisticalModel>

Parameters:

o1 - the first object

o2 - the second object

Returns:

the distance

Throws:

Exception - if the distance could not be computed

getPairwiseDistanceMatrix

public double[][] getPairwiseDistanceMatrix(int numThreads,
                                   StatisticalModel... objects)
                                     throws Exception

Multi-threaded computation of the pairwise distance matrix.

Parameters:

numThreads - the number of threads

objects - the models

Returns:

the distance matrix

Throws:

Exception - if the distance could not be computed