de.jstacs.sequenceScores

Interface QuickScanningSequenceScore

All Superinterfaces:

Cloneable, SequenceScore, Storable

All Known Implementing Classes:

BayesianNetworkDiffSM, LimitedSparseLocalInhomogeneousMixtureDiffSM_higherOrder, MarkovModelDiffSM

public interface QuickScanningSequenceScore
extends SequenceScore

Interface for SequenceScore that provide additional methods for computing scores of infix sequences and filtering infix sequences.

Author:

Jan Grau

Method Summary

Modifier and Type	Method and Description
void	fillInfixScore(int[] seq, int start, int length, double[] scores) Computes the position-wise scores of an infix of the sequence seq (which must be encoded by the same alphabet as this QuickScanningSequenceScore) beginning at start and extending for length positions.
boolean[][]	getInfixFilter(int kmer, double thresh, int... start) Computes arrays that indicate, for a given set of starting positions and a given k-mer length, if a sequence containing this k-mer may yield a score above threshold, choosing the best-scoring option among all non-specified positions (i.e., those outside the k-mer).

Methods inherited from interface de.jstacs.sequenceScores.SequenceScore

clone, getAlphabetContainer, getCharacteristics, getInstanceName, getLength, getLogScoreFor, getLogScoreFor, getLogScoreFor, getLogScoreFor, getLogScoreFor, getNumericalCharacteristics, isInitialized, toString

Methods inherited from interface de.jstacs.Storable

toXML

Method Detail

fillInfixScore

void fillInfixScore(int[] seq,
                    int start,
                    int length,
                    double[] scores)

Computes the position-wise scores of an infix of the sequence seq (which must be encoded by the same alphabet as this QuickScanningSequenceScore) beginning at start and extending for length positions. The scores are computed per position and filled into the provided array scores, which is of the same length as seq. This must be implemented such that ToolBox.sum(double...) applied to scores computed from 0 to seq.length returns the same value as SequenceScore.getLogScoreFor(de.jstacs.data.sequences.Sequence) called on the IntSequence created from seq.

Parameters:

seq - the sequence

start - the start of the infix

length - the length of the infix

scores - the array of scores to be (partly) filled

getInfixFilter

boolean[][] getInfixFilter(int kmer,
                           double thresh,
                           int... start)

Computes arrays that indicate, for a given set of starting positions and a given k-mer length, if a sequence containing this k-mer may yield a score above threshold, choosing the best-scoring option among all non-specified positions (i.e., those outside the k-mer). This method is implemented as an upper bound on the scores, i.e., there may be k-mers that are considered to score above threshold (i.e., that have entry true) although they are not, but there may not be k-mers that are considered to be below threshold (i.e., that have entry false), although there exist sequences containing this k-mer that do. The returned array is indexed by the starting positions (in the same order as provided in starts) in the first dimension, and in the second dimension it is indexed by an integer representation of the k-mers, assigning the highest priority to the first k-mer position, i.e.,

where denotes the size of the alphabet, the length of the k-mer (starting at 0 in this case), and denotes the function encoding symbols from the alphabet as integers (see DiscreteAlphabet).

Parameters:

kmer - the k-mer length

thresh - the threshold

start - the starting position(s)

Returns:

the indicator arrays