de.jstacs.motifDiscovery

Class SignificantMotifOccurrencesFinder

java.lang.Object

de.jstacs.motifDiscovery.SignificantMotifOccurrencesFinder

public class SignificantMotifOccurrencesFinder
extends Object

This class enables the user to predict motif occurrences given a specific significance level.

Author:

Jan Grau, Jens Keilwagen

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static interface	SignificantMotifOccurrencesFinder.JoinMethod Interface for methods that combine several profiles over the same sequence into one common profile
static class	SignificantMotifOccurrencesFinder.RandomSeqType
static class	SignificantMotifOccurrencesFinder.SumOfProbabilities Joins several profiles containing log-probabilities into one profile containing the logarithm of the sum of the probabilities of the single profiles.

Constructor Summary

Constructors

Constructor and Description
SignificantMotifOccurrencesFinder(MotifDiscoverer disc, DataSet bg, double[] weights, double sign) This constructor creates an instance of SignificantMotifOccurrencesFinder that uses a DataSet to determine the siginificance level.
SignificantMotifOccurrencesFinder(MotifDiscoverer disc, SignificantMotifOccurrencesFinder.JoinMethod joiner, DataSet bg, double[] weights, double sign) This constructor creates an instance of SignificantMotifOccurrencesFinder that uses a DataSet to determine the siginificance level.
SignificantMotifOccurrencesFinder(MotifDiscoverer disc, SignificantMotifOccurrencesFinder.RandomSeqType type, boolean oneHistogram, int numSequences, double sign) This constructor creates an instance of SignificantMotifOccurrencesFinder that uses the given SignificantMotifOccurrencesFinder.RandomSeqType to determine the siginificance level.
SignificantMotifOccurrencesFinder(MotifDiscoverer disc, SignificantMotifOccurrencesFinder.RandomSeqType type, SignificantMotifOccurrencesFinder.JoinMethod joiner, boolean oneHistogram, int numSequences, double sign) This constructor creates an instance of SignificantMotifOccurrencesFinder that uses the given SignificantMotifOccurrencesFinder.RandomSeqType to determine the siginificance level.

Method Summary

Modifier and Type	Method and Description
DataSet	annotateMotif(DataSet data, int motifIndex) This method annotates a DataSet.
DataSet	annotateMotif(DataSet data, int motifIndex, int addMax) This method annotates a DataSet.
DataSet	annotateMotif(int startPos, DataSet data, int motifIndex) This method annotates a DataSet starting in each sequence at startPos.
DataSet	annotateMotif(int startPos, DataSet data, int motifIndex, int addMax, boolean addAnnotation) This method annotates a DataSet starting in each sequence at startPos.
MotifAnnotation[]	findSignificantMotifOccurrences(int motif, Sequence seq, int start) This method finds the significant motif occurrences in the sequence.
MotifAnnotation[]	findSignificantMotifOccurrences(int motif, Sequence seq, int addMax, int start) This method finds the significant motif occurrences in the sequence.
DataSet	getBindingSites(DataSet data, int motifIndex) This method returns a DataSet containing the predicted binding sites.
DataSet	getBindingSites(int startPos, DataSet data, int motifIndex, int addMax, int addLeft, int addRight) This method returns a DataSet containing the predicted binding sites.
double	getFactorForAucPR() This method returns a factor that must be multiplied to scores for computing PR curves.
MotifDiscoverer	getMotifDiscoverer() This method returns a clone of the internally used MotifDiscoverer.
double	getNumberOfBoundSequences(DataSet data, double[] weights, int motifIndex) Returns the number of sequences in data that are predicted to be bound at least once by motif no.
double	getOffsetForAucPR() This method returns an offset that must be added to scores for computing PR curves.
double[][]	getPWM(int motif, DataSet data, double[] weights, int addLeft, int addRight) Returns the Position weight matrix (PWM) of the binding sites of motif motif in the data set data of the MotifDiscoverer of this SignificantMotifOccurrencesFinder.
Pair<double[][],double[]>	getPWMAndPosDist(int motif, DataSet data, double[] weights, double[] mean, int addLeft, int addRight) Returns the Position weight matrix (PWM) of the binding sites of motif motif in the data set data of the MotifDiscoverer of this SignificantMotifOccurrencesFinder together with standard deviation of binding site positions computed using the provided mean values for each sequence.
Pair<double[][],double[]>	getPWMAndPosDist(int motif, DataSet data, double[] weights, double[] mean, int addLeft, int addRight, LinkedList<Sequence> bs, DoubleList bsWeight, DoubleList bsScores) Returns the position weight matrix and standard deviation of the position distribution using the given mean.
Pair<double[][][],int[][]>	getPWMAndPositions(int motif, DataSet data, double[] weights, int addLeft, int addRight) Returns the Position weight matrix (PWM) of the binding sites of motif motif in the data set data of the MotifDiscoverer of this SignificantMotifOccurrencesFinder together with the positions of the binding sites within the sequences of data and the corresponding p-values.
protected double[][]	getPWMAndPositions(int motif, DataSet data, double[] weights, int addLeft, int addRight, int[][] positions, double[][] pvals, double[] mean, double[] sd, LinkedList<Sequence> bs, DoubleList bsWeights, DoubleList bsScores) Returns the Position weight matrix (PWM) of the binding sites of motif motif in the data set data of the MotifDiscoverer of this SignificantMotifOccurrencesFinder and fills with the positions of the binding sites within the sequences of data and the corresponding p-values into the corresponding arrays.
IntList	getStartPositions(int startPos, DataSet data, int motifIndex, int addMax) This method returns a list of start positions of binding sites.
double[][]	getValuesForEachNucleotide(DataSet data, int motif, boolean addOnlyBest) This method determines a score for each possible starting position in each of the sequences in data that this position is covered by at least one motif occurrence of the motif with index index.

Methods inherited from class java.lang.Object

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

Constructor Detail

SignificantMotifOccurrencesFinder

public SignificantMotifOccurrencesFinder(MotifDiscoverer disc,
                                         SignificantMotifOccurrencesFinder.RandomSeqType type,
                                         boolean oneHistogram,
                                         int numSequences,
                                         double sign)

This constructor creates an instance of SignificantMotifOccurrencesFinder that uses the given SignificantMotifOccurrencesFinder.RandomSeqType to determine the siginificance level.

Parameters:

disc - the MotifDiscoverer for the prediction

type - the type that determines how the significance level is determined

oneHistogram - a switch to decide whether to use one background distribution histogram for all sequence or sequence specific background distribution histograms

numSequences - the number of sampled sequence instances used to determine the significance level

sign - the significance level

SignificantMotifOccurrencesFinder

public SignificantMotifOccurrencesFinder(MotifDiscoverer disc,
                                         SignificantMotifOccurrencesFinder.RandomSeqType type,
                                         SignificantMotifOccurrencesFinder.JoinMethod joiner,
                                         boolean oneHistogram,
                                         int numSequences,
                                         double sign)

This constructor creates an instance of SignificantMotifOccurrencesFinder that uses the given SignificantMotifOccurrencesFinder.RandomSeqType to determine the siginificance level.

Parameters:

disc - the MotifDiscoverer for the prediction

type - the type that determines how the significance level is determined

joiner - the SignificantMotifOccurrencesFinder.JoinMethod that defines how the profiles of the same motif in different components shall be joined

oneHistogram - a switch to decide whether to use one background distribution histogram for all sequence or sequence specific background distribution histograms

numSequences - the number of sampled sequence instances used to determine the significance level

sign - the significance level

SignificantMotifOccurrencesFinder

public SignificantMotifOccurrencesFinder(MotifDiscoverer disc,
                                         DataSet bg,
                                         double[] weights,
                                         double sign)

This constructor creates an instance of SignificantMotifOccurrencesFinder that uses a DataSet to determine the siginificance level.

Parameters:

disc - the MotifDiscoverer for the prediction

bg - the background data set

weights - the weights of the background data set, can be null

sign - the significance level

SignificantMotifOccurrencesFinder

public SignificantMotifOccurrencesFinder(MotifDiscoverer disc,
                                         SignificantMotifOccurrencesFinder.JoinMethod joiner,
                                         DataSet bg,
                                         double[] weights,
                                         double sign)

This constructor creates an instance of SignificantMotifOccurrencesFinder that uses a DataSet to determine the siginificance level.

Parameters:

disc - the MotifDiscoverer for the prediction

joiner - the SignificantMotifOccurrencesFinder.JoinMethod that defines how the profiles of the same motif in different components shall be joined

bg - the background data set

weights - the weights of the background data set, can be null

sign - the significance level

Method Detail

findSignificantMotifOccurrences

public MotifAnnotation[] findSignificantMotifOccurrences(int motif,
                                                         Sequence seq,
                                                         int start)
                                                  throws Exception

This method finds the significant motif occurrences in the sequence.

Parameters:

motif - the motif index

seq - the sequence

start - the start position

Returns:

an array of MotifAnnotation for the sequence

Throws:

Exception - if the background sample could not be created, or some of the scores could not be computed

findSignificantMotifOccurrences

public MotifAnnotation[] findSignificantMotifOccurrences(int motif,
                                                         Sequence seq,
                                                         int addMax,
                                                         int start)
                                                  throws Exception

This method finds the significant motif occurrences in the sequence.

Parameters:

motif - the motif index

seq - the sequence

addMax - the number of motif occurrences that can at most be annotated

start - the start position

Returns:

an array of MotifAnnotation for the sequence

Throws:

Exception - if the background sample could not be created, or some of the scores could not be computed

getPWM

public double[][] getPWM(int motif,
                         DataSet data,
                         double[] weights,
                         int addLeft,
                         int addRight)
                  throws Exception

Returns the Position weight matrix (PWM) of the binding sites of motif motif in the data set data of the MotifDiscoverer of this SignificantMotifOccurrencesFinder.

Parameters:

motif - the index of the motif

data - the data set

weights - the weights on the individual sequences of the data set

addLeft - the number of positions to add to the left side of motif occurrences

addRight - the number of positions to add to the right side of motif occurrences

Returns:

the PWM

Throws:

Exception - if something went wrong

getPWMAndPositions

public Pair<double[][][],int[][]> getPWMAndPositions(int motif,
                                                     DataSet data,
                                                     double[] weights,
                                                     int addLeft,
                                                     int addRight)
                                              throws Exception

Returns the Position weight matrix (PWM) of the binding sites of motif motif in the data set data of the MotifDiscoverer of this SignificantMotifOccurrencesFinder together with the positions of the binding sites within the sequences of data and the corresponding p-values.

Parameters:

motif - the index of the motif

data - the data set

weights - the weights on the individual sequences of the data set

addLeft - the number of positions to add to the left side of motif occurrences

addRight - the number of positions to add to the right side of motif occurrences

Returns:

a Pair containing the PWM at index 0 of the first element, the p-values at index 1 of the first element, and the positions as second element. P-value and position arrays are indexed by the index of the corresponding sequence in data.

Throws:

Exception - if something went wrong

getPWMAndPositions

protected double[][] getPWMAndPositions(int motif,
                                        DataSet data,
                                        double[] weights,
                                        int addLeft,
                                        int addRight,
                                        int[][] positions,
                                        double[][] pvals,
                                        double[] mean,
                                        double[] sd,
                                        LinkedList<Sequence> bs,
                                        DoubleList bsWeights,
                                        DoubleList bsScores)
                                 throws Exception

Returns the Position weight matrix (PWM) of the binding sites of motif motif in the data set data of the MotifDiscoverer of this SignificantMotifOccurrencesFinder and fills with the positions of the binding sites within the sequences of data and the corresponding p-values into the corresponding arrays. Additionally, the standard deviation of binding site positions is computed using the provided mean values for each sequence.

Parameters:

motif - the index of the motif

data - the data set

weights - the weights on the individual sequences of the data set

addLeft - the number of positions to add to the left side of motif occurrences

addRight - the number of positions to add to the right side of motif occurrences

positions - the array filled with the positions. First dimension must contain as many entries as data has sequences. May be null.

pvals - the array filled with the p-values. First dimension must contain as many entries as data has sequences. May be null.

mean - the means for the individual sequences. May be null.

sd - array of lenght 1 filled with the determined standard deviation. May be null.

bs - the list of binding sites (filled if not null)

bsWeights - the list of weights of the sequences containing the binding sites (filled if not null)

bsScores - the list of binding site scores (filled if not null)

Returns:

the PWM

Throws:

Exception - if something went wrong

getPWMAndPosDist

public Pair<double[][],double[]> getPWMAndPosDist(int motif,
                                                  DataSet data,
                                                  double[] weights,
                                                  double[] mean,
                                                  int addLeft,
                                                  int addRight)
                                           throws Exception

Returns the Position weight matrix (PWM) of the binding sites of motif motif in the data set data of the MotifDiscoverer of this SignificantMotifOccurrencesFinder together with standard deviation of binding site positions computed using the provided mean values for each sequence.

Parameters:

motif - the index of the motif

data - the data set

weights - the weights on the individual sequences of the data set

mean - the means for the individual sequences. Must be as long as data has sequences.

addLeft - the number of positions to add to the left side of motif occurrences

addRight - the number of positions to add to the right side of motif occurrences

Returns:

a Pair containing the PWM as first element and standard deviation in array of length 1 as second element.

Throws:

Exception - if something went wrong

getPWMAndPosDist

public Pair<double[][],double[]> getPWMAndPosDist(int motif,
                                                  DataSet data,
                                                  double[] weights,
                                                  double[] mean,
                                                  int addLeft,
                                                  int addRight,
                                                  LinkedList<Sequence> bs,
                                                  DoubleList bsWeight,
                                                  DoubleList bsScores)
                                           throws Exception

Returns the position weight matrix and standard deviation of the position distribution using the given mean.

Parameters:

motif - the index of the motif

data - the DataSet in which binding sites are predicted

weights - the weights on the sequences of data

mean - the mean of the position distribution

addLeft - the number of positions to add to the left flank of the motif

addRight - the number of positions to add to the right flank of the motif

bs - the list of binding sites, filled if not null

bsWeight - the list of weights of the sequences containing the binding sites, filled if not null

bsScores - the scores of the predicted binding sites, filled if not null

Returns:

the PWM and standard deviation (in array element 0)

Throws:

Exception - if something went wrong

annotateMotif

public DataSet annotateMotif(DataSet data,
                             int motifIndex)
                      throws Exception

This method annotates a DataSet.

Parameters:

data - the DataSet

motifIndex - the index of the motif

Returns:

an annotated DataSet

Throws:

Exception - if something went wrong

See Also:

annotateMotif(int, DataSet, int)

annotateMotif

public DataSet annotateMotif(int startPos,
                             DataSet data,
                             int motifIndex)
                      throws Exception

This method annotates a DataSet starting in each sequence at startPos.

Parameters:

startPos - the start position used for all sequences

data - the DataSet

motifIndex - the index of the motif

Returns:

an annotated DataSet

Throws:

Exception - if something went wrong

See Also:

annotateMotif(int, DataSet, int)

annotateMotif

public DataSet annotateMotif(DataSet data,
                             int motifIndex,
                             int addMax)
                      throws Exception

This method annotates a DataSet. At most, addMax motif occurrences of the motif instance will be annotated.

Parameters:

data - the DataSet

motifIndex - the index of the motif

addMax - the number of motif occurrences that can at most be annotated for each motif instance

Returns:

an annotated DataSet

Throws:

Exception - if something went wrong

See Also:

annotateMotif(int, DataSet, int)

annotateMotif

public DataSet annotateMotif(int startPos,
                             DataSet data,
                             int motifIndex,
                             int addMax,
                             boolean addAnnotation)
                      throws Exception

This method annotates a DataSet starting in each sequence at startPos. At most, addMax motif occurrences of the motif instance will be annotated.

Parameters:

startPos - the start position used for all sequences

data - the DataSet

motifIndex - the index of the motif

addMax - the number of motif occurrences that can at most be annotated for each motif instance

addAnnotation - a switch whether to add or replace the current annotation

Returns:

an annotated DataSet

Throws:

Exception - if something went wrong

See Also:

annotateMotif(int, DataSet, int)

getBindingSites

public DataSet getBindingSites(DataSet data,
                               int motifIndex)
                        throws Exception

This method returns a DataSet containing the predicted binding sites.

Parameters:

data - the DataSet

motifIndex - the index of the motif

Returns:

a DataSet containing the predicted binding sites

Throws:

Exception - if something went wrong

getBindingSites

public DataSet getBindingSites(int startPos,
                               DataSet data,
                               int motifIndex,
                               int addMax,
                               int addLeft,
                               int addRight)
                        throws Exception

This method returns a DataSet containing the predicted binding sites.

Parameters:

startPos - the start position used for all sequences

data - the DataSet

motifIndex - the index of the motif

addMax - the number of motif occurrences that can at most be annotated for each motif instance

addLeft - number of positions added to the left of the predicted motif occurrence

addRight - number of positions added to the right of the predicted motif occurrence

Returns:

a DataSet containing the predicted binding sites

Throws:

Exception - if something went wrong

getStartPositions

public IntList getStartPositions(int startPos,
                                 DataSet data,
                                 int motifIndex,
                                 int addMax)
                          throws Exception

This method returns a list of start positions of binding sites.

Parameters:

startPos - the start position used for all sequences

data - the DataSet

motifIndex - the index of the motif

addMax - the number of motif occurrences that can at most be annotated for each motif instance

Returns:

a list of start positions

Throws:

Exception - if something went wrong

getNumberOfBoundSequences

public double getNumberOfBoundSequences(DataSet data,
                                        double[] weights,
                                        int motifIndex)
                                 throws Exception

Returns the number of sequences in data that are predicted to be bound at least once by motif no. motif.

Parameters:

data - the data

weights - the weights of the data

motifIndex - the index of the motif

Returns:

the number of sequences in data bound by motif motif

Throws:

Exception - if the background sample for the prediction could not be created or some of the scores could not be computed

getOffsetForAucPR

public double getOffsetForAucPR()

This method returns an offset that must be added to scores for computing PR curves. If this SignificantMotifOccurrencesFinder was instantiated using oneHistogram=true, the getValuesForEachNucleotide(DataSet, int, boolean) returns scores and no offset is needed. Otherwise, it returns p-values and, hence, 1-(p-value) must be used for the PR curve and the offset is 1.

Returns:

the offset

See Also:

MotifDiscoveryAssessment.getSortedValuesForMotifAndFlanking(DataSet, double[][], double, double, String)

getFactorForAucPR

public double getFactorForAucPR()

This method returns a factor that must be multiplied to scores for computing PR curves. If this SignificantMotifOccurrencesFinder was instantiated using oneHistogram=true, the getValuesForEachNucleotide(DataSet, int, boolean) returns scores and a factor of 1 is appropriate. Otherwise, it returns p-values and, hence, 1-(p-value) must be used for the PR curve and the factor is -1.

Returns:

the factor

See Also:

MotifDiscoveryAssessment.getSortedValuesForMotifAndFlanking(DataSet, double[][], double, double, String)

getValuesForEachNucleotide

public double[][] getValuesForEachNucleotide(DataSet data,
                                             int motif,
                                             boolean addOnlyBest)
                                      throws Exception

This method determines a score for each possible starting position in each of the sequences in data that this position is covered by at least one motif occurrence of the motif with index index. If the SignificantMotifOccurrencesFinder was constructed using oneHistogram=true the returned values are arbitrary scores, and p-values otherwise.

Parameters:

data - the DataSet

motif - the motif index

addOnlyBest - a switch whether to add only the best

Returns:

an array containing for each sequence an array with the scores for each starting position in the sequence

Throws:

Exception - if something went wrong during the computation of the scores of the MotifDiscoverer

See Also:

MotifDiscoveryAssessment.getSortedValuesForMotifAndFlanking(DataSet, double[][], double, double, String), getOffsetForAucPR(), getFactorForAucPR()

getMotifDiscoverer

public MotifDiscoverer getMotifDiscoverer()
                                   throws CloneNotSupportedException

This method returns a clone of the internally used MotifDiscoverer.

Returns:

clone of the internally used MotifDiscoverer

Throws:

CloneNotSupportedException - if the MotifDiscoverer can not be cloned correctly