StochHMM::sequences Class Reference

#include <sequences.h>

List of all members.

Public Member Functions
	sequences ()
	sequences (size_t sz)
	sequences (tracks *tr)
	sequences (const sequences &)
sequences &	operator= (const sequences &)
	Assignment Operator.
	~sequences ()
	Destroy sequences.
double	realValue (int, size_t)
double	realValue (size_t, size_t)
short	seqValue (int, size_t)
sequence *	getSeq (size_t)
double	getAttrib (size_t iter)
std::string &	getHeader ()
std::string &	getHeader (size_t iter)
std::string *	getUndigitized (size_t iter)
size_t	size ()
size_t	getLength (size_t iter)
size_t	getLength ()
bool	exDefDefined (size_t)
bool	exDefDefined (size_t, size_t)
double	getWeight (size_t, size_t)
bool	exDefDefined ()
	Get Weight value for state at position.
void	print ()
	Is External definition defined.
std::string	stringify ()
std::string	undigitize ()
	Get string of sequences.
void	addSeq (sequence *sq)
	Get sequence based on alphabet.
void	addSeq (sequence *, size_t)
void	addSeq (sequence *, track *)
	Add sequence for track.
void	setLength (size_t len)
	Set Length of sequences.
void	setExDef (ExDefSequence *ex)
bool	isSameSize ()
sequence &	operator[] (size_t index)
void	getFastas (const std::string &, track *)

Private Attributes
ExDefSequence *	external
std::vector< sequence * >	seq
size_t	length
size_t	num_of_sequences
bool	related_sequences
bool	same_length

---

Detailed Description

Container hold the track sequence(s) Each sequence have to be the same length Created to pass multiple sequence tracks of different datasets to the HMM If you need multiple unrelated sequences, use std::vector<sequence> instead.

Definition at line 51 of file sequences.h.

---

Constructor & Destructor Documentation

StochHMM::sequences::sequences

(

)

Definition at line 35 of file sequences.cpp.

References external, length, num_of_sequences, related_sequences, and same_length.

                        {
        length=-1;
        external=NULL;
        related_sequences=false;
        num_of_sequences=0;
        same_length=true;
    }

StochHMM::sequences::sequences

(

size_t

sz

)

Create an empty sequences

Parameters:

sz	Number of sequences

Definition at line 46 of file sequences.cpp.

References external, length, num_of_sequences, related_sequences, and same_length.

                                 :seq(sz,NULL){
        length=-1;
        external=NULL;
        related_sequences=true;
        num_of_sequences=sz;
        same_length=true;
    }

StochHMM::sequences::sequences

(

tracks *

tr

)

Create a sequences data t

Parameters:

sz	Number of sequences

Definition at line 57 of file sequences.cpp.

References external, length, num_of_sequences, related_sequences, same_length, and StochHMM::tracks::size().

                                  :seq(tr->size(),NULL){
        length=-1;
        external=NULL;
        related_sequences=true;
        num_of_sequences=tr->size();
        same_length=true;
    }

StochHMM::sequences::sequences

(

const sequences &

rhs

)

Definition at line 66 of file sequences.cpp.

References external, length, num_of_sequences, related_sequences, same_length, and seq.

                                            {
        external = (rhs.external==NULL) ? NULL : new(std::nothrow) ExDefSequence(*rhs.external);
      
        if (rhs.external != NULL && external==NULL){
            std::cerr << "OUT OF MEMORY\nFile" << __FILE__ << "Line:\t"<< __LINE__ << std::endl;
            exit(1);
        }
        
        length = rhs.length;
        same_length=rhs.same_length;
        num_of_sequences=rhs.num_of_sequences;
        related_sequences=rhs.related_sequences;
        
        for(size_t i=0;i<seq.size();i++){
            sequence* temp=NULL;
            if (rhs.seq[i]!=NULL){
                temp = new(std::nothrow) sequence(*rhs.seq[i]);
                
                if (temp == NULL){
                    std::cerr << "OUT OF MEMORY\nFile" << __FILE__ << "Line:\t"<< __LINE__ << std::endl;
                    exit(1);
                }
            }            
            seq.push_back(temp);
        }
        
    }

StochHMM::sequences::~sequences

(

)

Destroy sequences.

Definition at line 96 of file sequences.cpp.

References external, and seq.

                         {
        for(size_t i=0;i<seq.size();i++){
            delete seq[i];
            seq[i]=NULL;
        }
        delete external;
        external = NULL;
    }

---

Member Function Documentation

void StochHMM::sequences::addSeq

(

sequence *

sq

)

Get sequence based on alphabet.

Add a sequence to the sequences *Sequence is added to a certain position based on the track used by the sequence

Parameters:

sq	Pointer to sequence

Definition at line 262 of file sequences.cpp.

References StochHMM::track::getIndex(), StochHMM::sequence::getLength(), StochHMM::sequence::getTrack(), num_of_sequences, related_sequences, seq, and setLength().

Referenced by addSeq(), getFastas(), and StochHMM::seqTracks::getNext().

                                      {
        if (related_sequences){
            track* trk = sq->getTrack();
            seq[trk->getIndex()]=sq;
            setLength(sq->getLength());
            return;
        }
        else{
            seq.push_back(sq);
            setLength(sq->getLength());
            num_of_sequences++;
        }
        
    }

void StochHMM::sequences::addSeq	(	sequence *	sq,
		size_t	iter
	)

Add sequence in the track position

Parameters:

Add	a sequence to the sequences at a certain position *Sequence is added to a certain position based on the iterator
sq	Pointer to sequence
iter	Position in sequences to add sequence

Definition at line 281 of file sequences.cpp.

References StochHMM::sequence::getLength(), num_of_sequences, related_sequences, seq, and setLength().

                                                   {
        if (seq.size()>iter){  //If position already exists just add it
            seq[iter]=sq;
        }
        else{ //If the position doesn't exist extend the vector, then add
            seq.resize(iter+1,NULL);
            seq[iter]=sq;
        }
        
        setLength(sq->getLength());
        
        if (!related_sequences){
            num_of_sequences=iter;
        }
        
        return;
    }

void StochHMM::sequences::addSeq	(	sequence *	sq,
		track *	tr
	)

Add sequence for track.

Add a sequence to the sequences given a certain track *Sequence is added to a certain position based on the track

Parameters:

sq	Pointer to sequence
tr	Track to use when adding sequence

Definition at line 303 of file sequences.cpp.

References addSeq(), StochHMM::track::getIndex(), and related_sequences.

                                                {
        if (tr!=NULL){
            if (related_sequences){
                size_t index = tr->getIndex();
                addSeq(sq,index);
            }
            else{
                addSeq(sq);
            }
        }
        else{
            if (related_sequences){
                std::cerr << "Track is undefined for related sequences.  Unable to add related sequences if track is not defined\n";
                exit(1);
            }
            else{
                addSeq(sq);
            }
        }
        return;
    }

bool StochHMM::sequences::exDefDefined

(

size_t

pos

)

Check to see if ther is an external definition defined for a certain position within the sequences

Parameters:

pos	Position within the sequence

Returns:

True if there is an external definition defined for the position

Definition at line 223 of file sequences.cpp.

References StochHMM::ExDefSequence::defs, and external.

Referenced by StochHMM::trellis::backward(), StochHMM::trellis::fast_complex_viterbi(), StochHMM::trellis::forward(), StochHMM::trellis::naive_nth_viterbi(), StochHMM::trellis::naive_stochastic_viterbi(), StochHMM::trellis::naive_viterbi(), StochHMM::trellis::posterior(), StochHMM::trellis::simple_backward(), StochHMM::trellis::simple_forward(), StochHMM::trellis::simple_nth_viterbi(), StochHMM::trellis::simple_posterior(), StochHMM::trellis::simple_simple_stochastic_viterbi(), StochHMM::trellis::simple_stochastic_forward(), StochHMM::trellis::simple_stochastic_viterbi(), StochHMM::trellis::simple_viterbi(), StochHMM::trellis::sparse_complex_viterbi(), StochHMM::trellis::stochastic_forward(), StochHMM::trellis::stochastic_viterbi(), StochHMM::trellis::trellis(), and StochHMM::trellis::viterbi().

                                          {
        if (external && external->defs[pos]){
            return true;
        }
        else{
            return false;
        }
    }

bool StochHMM::sequences::exDefDefined	(	size_t	pos,
		size_t	stateIter
	)

Check to see if there is an external definition defined for a certain state at a certain position

Parameters:

pos	Position within the sequence

Returns:

True if there is an external definition fo the the state and position

Definition at line 235 of file sequences.cpp.

References StochHMM::ExDefSequence::defined(), StochHMM::ExDefSequence::defs, and external.

                                                            {
        if (external && external->defs[pos]){
            if (external->defined(stateIter)){
                return true;
            }
        }
        return false;
    }

bool StochHMM::sequences::exDefDefined

(

)

Get Weight value for state at position.

Check to see if there are any external definitions defined

Returns:

true if there are external definitions defined

Definition at line 247 of file sequences.cpp.

References external.

                                {
        return external;
    }

double StochHMM::sequences::getAttrib ( size_t  iter )

inline

Get the attribute value for a particular sequence

Parameters:

iter	Sequence to get the attribute from

Returns:

double value of the attribute set for the sequence

Definition at line 79 of file sequences.h.

References num_of_sequences, and seq.

                                            {
            if (iter<num_of_sequences && seq[iter]!=NULL){
                return seq[iter]->getAttrib();  //Returns attrib value for Track (i)
            }
            std::cerr << "No sequence defined at iterator " << iter << std::endl;
            exit(1);
        }

void StochHMM::sequences::getFastas	(	const std::string &	filename,
		track *	tr
	)

Definition at line 347 of file sequences.cpp.

References addSeq(), and StochHMM::sequence::getFasta().

Referenced by main().

                                                                     {
                std::ifstream file;
                file.open(filename.c_str());
                if (!file.is_open()){
                        std::cerr << "Couln't open the file:" << filename << "\n";
                        exit(2);
                }
                while(!file.eof()){
                        sequence* sq = new(std::nothrow) sequence(false);
                        
                        if (sq==NULL){
                                std::cerr << "OUT OF MEMORY\nFile" << __FILE__ << "Line:\t"<< __LINE__ << std::endl;
                                exit(1);
                        }
                        bool success = sq->getFasta(file, tr);
                        
                        if (!success){
                                delete sq;
                        }
                        addSeq(sq,tr);
                }
                return;
        }

std::string& StochHMM::sequences::getHeader ( )

inline

Get the header for the first sequence

Returns:

std::string& The header for the first sequence in sequences

Definition at line 90 of file sequences.h.

References num_of_sequences, and seq.

Referenced by perform_nbest_decoding(), perform_posterior(), perform_stochastic_decoding(), perform_viterbi_decoding(), print_limited_posterior(), and print_posterior().

                                     {
            if (num_of_sequences>0 && seq[0]!=NULL){
                return seq[0]->header;
            }
            std::cerr << "No sequence defined." << std::endl;
            exit(1);
        } 

std::string& StochHMM::sequences::getHeader ( size_t  iter )

inline

Get the header for the ith sequence

Parameters:

iter	size_t iterator for ith sequence

Returns:

std::string& The header for the ith sequence

Definition at line 102 of file sequences.h.

References num_of_sequences, and seq.

                                                {
            if (iter<num_of_sequences){
                if (seq[iter]!=NULL){
                    return seq[iter]->header;
                }
            }
            std::cerr << "No sequence defined at iterator " << iter << std::endl;
            exit(1);
        }

size_t StochHMM::sequences::getLength ( size_t  iter )

inline

Get the length of the ith sequence

Parameters:

iter	size_t iterator

Returns:

size_t value of length of sequence at iter

Definition at line 135 of file sequences.h.

References num_of_sequences, and seq.

Referenced by StochHMM::trellis::backward(), StochHMM::trellis::forward(), StochHMM::seqTracks::getNext(), StochHMM::trellis::naive_stochastic_viterbi(), StochHMM::trellis::naive_viterbi(), StochHMM::trellis::posterior(), StochHMM::trellis::simple_backward(), StochHMM::trellis::simple_forward(), StochHMM::trellis::simple_posterior(), StochHMM::trellis::simple_simple_stochastic_viterbi(), StochHMM::trellis::simple_stochastic_viterbi(), StochHMM::trellis::simple_viterbi(), StochHMM::trellis::stochastic_forward(), StochHMM::trellis::stochastic_viterbi(), StochHMM::trellis::trellis(), and StochHMM::trellis::viterbi().

                                            {
            if (iter<num_of_sequences && seq[iter]!=NULL){
                return seq[iter]->getLength(); //Get lenght of sequence in position (i)
            }
            return 0;
        } 

size_t StochHMM::sequences::getLength ( )

inline

Get the length of sequences in general *All of the sequence(s) should be the same length

Returns:

size_t value of length of all sequences

Definition at line 145 of file sequences.h.

References length.

                                 {
                        return length;
//            if (related_sequences){
//                return length; // Get length of related sequence
//            }
//            return 0;
        };

sequence * StochHMM::sequences::getSeq

(

size_t

iter

)

Get pointer to ith sequence from sequences

Parameters:

iter	Iterator to use for extracting sequence;

Returns:

sequence* pointer to sequence

NULL if no sequence exists at iter

Definition at line 174 of file sequences.cpp.

References seq.

Referenced by StochHMM::PWM::scoreSimple(), StochHMM::PWM::scoreUndefSpacer(), StochHMM::PWM::scoreVariableSpacer(), and StochHMM::trellis::transitionFuncTraceback().

                                          {
        if(iter<seq.size()){
            return seq[iter];
        }
        
        return NULL;
    }

std::string* StochHMM::sequences::getUndigitized ( size_t  iter )

inline

Get the undigitized ith sequence from sequences

Parameters:

iter	size_t iterator for ith sequence

Returns:

std::string of undigitized sequence at ith position

Definition at line 116 of file sequences.h.

References seq.

Referenced by StochHMM::emissionFuncParam::evaluate(), StochHMM::seqTracks::getNext(), and StochHMM::trellis::transitionFuncTraceback().

                                                     {
            if (iter>seq.size()){
                std::cerr << "getUndigitized(size_t iter) called where iter is out of range\n";
                return NULL;
            }
            else{
                return seq[iter]->getUndigitized();
            }
        };

double StochHMM::sequences::getWeight	(	size_t	position,
		size_t	stateIter
	)

Get the weight for the state at a certain position in the sequence

Parameters:

position	Position with the sequence
stateIter	integer iterator of the state

Definition at line 254 of file sequences.cpp.

References external, and StochHMM::ExDefSequence::getWeight().

Referenced by StochHMM::trellis::fast_complex_viterbi(), StochHMM::trellis::naive_forward(), StochHMM::trellis::naive_nth_viterbi(), StochHMM::trellis::naive_stochastic_forward(), StochHMM::trellis::naive_stochastic_viterbi(), StochHMM::trellis::naive_viterbi(), StochHMM::trellis::simple_backward(), StochHMM::trellis::simple_forward(), StochHMM::trellis::simple_nth_viterbi(), StochHMM::trellis::simple_posterior(), StochHMM::trellis::simple_simple_stochastic_viterbi(), StochHMM::trellis::simple_stochastic_forward(), StochHMM::trellis::simple_stochastic_viterbi(), StochHMM::trellis::simple_viterbi(), and StochHMM::trellis::sparse_complex_viterbi().

                                                                {
        return external->getWeight(position, stateIter);
    }

bool StochHMM::sequences::isSameSize ( )

inline

Definition at line 185 of file sequences.h.

References same_length.

                                {
            return same_length;
        }

sequences & StochHMM::sequences::operator=

(

const sequences &

rhs

)

Assignment Operator.

Definition at line 107 of file sequences.cpp.

References external, length, num_of_sequences, related_sequences, same_length, and seq.

                                                        {
        external = (rhs.external==NULL) ? NULL : new(std::nothrow) ExDefSequence(*rhs.external);
        
        if (rhs.external != NULL && external==NULL){
            std::cerr << "OUT OF MEMORY\nFile" << __FILE__ << "Line:\t"<< __LINE__ << std::endl;
            exit(1);
        }
        
        length = rhs.length;
        same_length=rhs.same_length;
        num_of_sequences=rhs.num_of_sequences;
        related_sequences=rhs.related_sequences;
        
        for(size_t i=0;i<seq.size();i++){
            sequence* temp=NULL;
            if (rhs.seq[i]!=NULL){
                temp = new(std::nothrow) sequence(*rhs.seq[i]);
                
                if (temp == NULL){
                    std::cerr << "OUT OF MEMORY\nFile" << __FILE__ << "Line:\t"<< __LINE__ << std::endl;
                    exit(1);
                }
            }            
            seq.push_back(temp);
        }
        
        return *this;
    }

sequence& StochHMM::sequences::operator[] ( size_t  index )

inline

Definition at line 189 of file sequences.h.

References seq.

{return *seq[index];}

void StochHMM::sequences::print ( )

inline

Is External definition defined.

Print the string representation of digitized sequencs to the stdout

Definition at line 160 of file sequences.h.

References stringify().

Referenced by main().

{std::cout<< stringify() << std::endl;}; //Print sequences to stdout

double StochHMM::sequences::realValue	(	int	trck,
		size_t	position
	)

Get the value from a real Number sequence for track trck at position

Parameters:

trck	Sequence track to use
position	Position in sequence to get value from

Returns:

double value of the real sequence at the position

Definition at line 141 of file sequences.cpp.

References seq.

Referenced by StochHMM::emm::get_emission().

                                                       {
        if (seq[trck]->realSeq){
            return seq[trck]->realValue(position);
        }
        return -INFINITY;
        
    }

double StochHMM::sequences::realValue	(	size_t	trck,
		size_t	position
	)

Get the value from a real Number sequence for track trck at position

Parameters:

trck	Sequence track to use
position	Position in sequence to get value from

Returns:

double value of the real sequence at the position

Definition at line 153 of file sequences.cpp.

References seq.

                                                          {
        if (seq[trck]->realSeq){
            return seq[trck]->realValue(position);
        }
        return -INFINITY;
        
    }

short StochHMM::sequences::seqValue	(	int	trck,
		size_t	position
	)

Get the digitized value from the sequence at trck at position

Parameters:

trck	Sequence track to use
position	Position in sequence to get the value from

Returns:

short digitized value of the sequence based on track type

Definition at line 166 of file sequences.cpp.

References seq.

                                                      {
        return seq[trck]->seqValue(position);
    }

void StochHMM::sequences::setExDef ( ExDefSequence *  ex )

inline

Set external definition

Parameters:

ex	External definition to assign to the sequences

Definition at line 183 of file sequences.h.

References external.

Referenced by StochHMM::seqTracks::getNext().

{external=ex;};

void StochHMM::sequences::setLength

(

size_t

len

)

Set Length of sequences.

Set the length of the sequence(s) in sequence Because all the sequences should be the same size If there size differs when adding a sequence

Exceptions:

sDifferentSizeSequences

thrown if the sizes differ

Definition at line 329 of file sequences.cpp.

References length, StochHMM::max(), related_sequences, and same_length.

Referenced by addSeq(), and StochHMM::seqTracks::getNext().

                                       {
        if (length == std::numeric_limits<size_t>::max()){
            length=len;
        }
        
        if (len==length){
            return;
        }
        else{
            same_length=false;
            if (related_sequences){
                std::cerr << "Sequences have different lengths. Sequences should all have the same length because they are suppose to be related (from different datasets).  For multiple unrelated sequence types, use a different structure.\n";
                exit(20);
            }
        }
        return;
    }

size_t StochHMM::sequences::size ( void  )

inline

Get the number of sequence type in sequences

Returns:

size_t value of size

Definition at line 130 of file sequences.h.

References num_of_sequences.

Referenced by StochHMM::PWM::scoreSimple(), StochHMM::PWM::scoreUndefSpacer(), StochHMM::PWM::scoreVariableSpacer(), stringify(), and undigitize().

{return num_of_sequences;} //Get number of sequences

std::string StochHMM::sequences::stringify

(

)

Get std:string representation of all the digitized sequence(s) in sequences

Returns:

std::string representation of all string (digitized)

Definition at line 184 of file sequences.cpp.

References StochHMM::track::getName(), StochHMM::int_to_string(), seq, and size().

Referenced by print().

                                  {
        std::string tmp;
        for(size_t i=0; i<size(); i++){
            
            
            if (seq[i]==NULL){
                tmp+= ">TRACK: " + int_to_string(i) + ":\t" ;
                tmp+= "<<EMPTY>>\n" ;
            }
            else{
                track* trk = seq[i]->getTrack();
                tmp+= ">" + trk->getName();
                tmp+= seq[i]->stringify();
            }
        }
        return tmp;
    }

std::string StochHMM::sequences::undigitize

(

)

Get string of sequences.

Get std::string representation of all the undigitized sequence(s) in sequences

Returns:

std::string representation of all string (digitized)

Definition at line 204 of file sequences.cpp.

References StochHMM::track::getName(), StochHMM::int_to_string(), seq, and size().

                                   {
        std::string output;
        for(size_t i=0;i<size();i++){
            if (seq[i]==NULL){
                output+= ">TRACK: " + int_to_string(i) + ":\t" ;
                output+= "<<EMPTY>>\n" ;
            }
            else{
                track* trk = seq[i]->getTrack();
                output+= ">" + trk->getName();
                output+= seq[i]->undigitize();
            }
        }
        return output;
    }

---

Member Data Documentation

ExDefSequence* StochHMM::sequences::external

private

Definition at line 195 of file sequences.h.

Referenced by exDefDefined(), getWeight(), operator=(), sequences(), setExDef(), and ~sequences().

size_t StochHMM::sequences::length

private

Definition at line 199 of file sequences.h.

Referenced by getLength(), operator=(), sequences(), and setLength().

size_t StochHMM::sequences::num_of_sequences

private

Definition at line 200 of file sequences.h.

Referenced by addSeq(), getAttrib(), getHeader(), getLength(), operator=(), sequences(), and size().

bool StochHMM::sequences::related_sequences

private

Definition at line 202 of file sequences.h.

Referenced by addSeq(), operator=(), sequences(), and setLength().

bool StochHMM::sequences::same_length

private

Definition at line 203 of file sequences.h.

Referenced by isSameSize(), operator=(), sequences(), and setLength().

std::vector<sequence*> StochHMM::sequences::seq

private

Definition at line 197 of file sequences.h.

Referenced by addSeq(), getAttrib(), getHeader(), getLength(), getSeq(), getUndigitized(), operator=(), operator[](), realValue(), sequences(), seqValue(), stringify(), undigitize(), and ~sequences().

---

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

• sequences.h
• sequences.cpp