SequenceReader/SequenceEncoder.h File Reference

#include "GlobalDefinitions.h"

Include dependency graph for SequenceEncoder.h:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Classes
class	SequenceEncoder
class	SequenceEncoderDNA
class	SequenceEncoderProtein
class	SequenceEncoderCodon
Typedefs
typedef Word	TranslationTable [numPossibleChars]
typedef Word	ExpandedTranslationTable [numPossibleChars *numPossibleChars]
typedef vector< char >	CodonList
Functions
static const int	numPossibleChars (1<< 8)
static const Word	nv (0xFF)
static const Word	firstCharInvalid (1<< 31)
static const Word	secondCharInvalid (1<< 30)
static const Word	someCharInvalid (firstCharInvalid|secondCharInvalid)
static const Word	maskBase (0x3)
static const Word	mask2Bases ((maskBase<< gBaseBits)|maskBase)
static const Word	maskCodon ((mask2Bases<< gBaseBits)|maskBase)
static const char	flaggedChar ('^')
ostream &	operator<< (ostream &os, CodonList &c)
void	codonize (const WordSequence &in, CodonList &codons, int readingFrame)
template<int CODON_SHIFT, int BASE_SHIFT>
Word	getCodonFromWord (const Word &w)
void	codonizeAndFlag (const WordSequence &in, CodonList &codons, int readingFrame)
void	codonizeAndFlagReverse (const WordSequence &in, CodonList &codons, int readingFrame)
void	print (const TranslationTable &tt)
Variables
static const TranslationTable	ttDNA
static const TranslationTable	ttProtein
static const TranslationTable	ttCodon
static const TranslationTable	ttCodonReverse

---

Typedef Documentation

typedef vector<char> CodonList

Definition at line 326 of file SequenceEncoder.h.

typedef Word ExpandedTranslationTable[numPossibleChars *numPossibleChars]

Definition at line 49 of file SequenceEncoder.h.

typedef Word TranslationTable[numPossibleChars]

Definition at line 48 of file SequenceEncoder.h.

---

Function Documentation

void codonize	(	const WordSequence &	in,
		CodonList &	codons,
		int	readingFrame
	)

Definition at line 340 of file SequenceEncoder.cpp.

References gBaseBits, gBasesPerCodon, gBitsPerWord, gCodonBits, WordSequence::getNumBasesInLast(), and gMaxBasesPerWord.

Referenced by MatchAlignerTranslated::createAlignment(), main(), HashTablePackedProtein::matchSequenceTranslatedDNA(), and HashTableTranslated::matchSequenceTranslatedDNA().

  {
    //    cout << "Codonizing for reading frame " << readingFrame << endl;
    static const Word mask((1<<gCodonBits)-1);

    int shift( gBitsPerWord-gCodonBits-gBaseBits*readingFrame );


    WordSequence::const_iterator i(in.begin());
    //    vector<char> codons;
    int codonsSoFar(0);
    int totalCodons
      ( (gMaxBasesPerWord*(in.size()-1) 
          + in.getNumBasesInLast() - readingFrame)/gBasesPerCodon );

    Word thisCodon;

    while( 1 )
    {
    
      if (shift>=0)
      {
        thisCodon = 0;
      }
      else
      {
        thisCodon = (*i<<-shift);
        i++;
        shift += gBitsPerWord;
      } 
      thisCodon |= (*i>>shift);
      thisCodon &= mask;

      codons.push_back(thisCodon);
      if (++codonsSoFar==totalCodons) break;
      if (shift == 0) { shift = gBitsPerWord; i++; }
      shift -= gCodonBits;

    } // ~while
  
    //    pEncoder_->encode( codons );


  } // ~SequenceReaderCodon::codonize

Here is the call graph for this function:

Here is the caller graph for this function:

void codonizeAndFlag	(	const WordSequence &	in,
		CodonList &	codons,
		int	readingFrame
	)

Definition at line 402 of file SequenceEncoder.cpp.

References flaggedChar(), gBaseBits, gCodonBits, gCursedWord, WordSequence::getNumBasesInLast(), gMaxBasesPerWord, gNumReadingFrames, mask2Bases(), and maskBase().

Referenced by HashTableTranslated::countWordsAndGetNames(), HashTableTranslated::hashAllWords(), and main().

{

  static const int DNAWordSizeForHashing(gMaxBasesPerWord-1);

  int codonsInLast((in.getNumBasesInLast()-readingFrame)/gNumReadingFrames);

  //  cout << "blx: " << in.size() << " " << in.getNumBasesInLast() << " " << codonsInLast << endl;
  int numCodons
    (   ( (in.size() == 0) 
          ? 0
          : ((in.size()-1)*DNAWordSizeForHashing)+in.getNumBasesInLast()
          -readingFrame )
      / gNumReadingFrames );

  codons.resize(numCodons);
  char* pCodon(static_cast<char*>(&(*codons.begin())));
  char* pEnd(static_cast<char*>(&(*codons.end())));
  WordSequence::const_iterator i(in.begin());
  WordSequence::const_iterator lastWord(in.end()-1);
  Word toCarry(~0);
  Word lastWordFlag(0);

  //  const Word maskBase(0x3);
  //  const Word maskCodon((0x3>>4)|(0x3>>2)|(0x3>>4));
  

  if (readingFrame==0)
  {
    for( ; i!=lastWord ; ++i )
    {

      *(pCodon++) = getCodonFromWord<4,0>(*i);
      *(pCodon++) = getCodonFromWord<3,0>(*i);
      *(pCodon++) = getCodonFromWord<2,0>(*i);
      *(pCodon++) = getCodonFromWord<1,0>(*i);
      *(pCodon++) = getCodonFromWord<0,0>(*i);

    } // ~for

    if (codonsInLast>=1)
      *(pCodon++) = getCodonFromWord<4,0>(*i);
    if (codonsInLast>=2)
      *(pCodon++) = getCodonFromWord<3,0>(*i);
    if (codonsInLast>=3)
      *(pCodon++) = getCodonFromWord<2,0>(*i);
    if (codonsInLast==4)
      *(pCodon++) = getCodonFromWord<1,0>(*i);
  } // ~if 
  else if (readingFrame==1)
  {

    for( ; i!=lastWord ; ++i )
    {
      if (toCarry!=~0)
      (*pCodon++) = ( ((*i)&gCursedWord)|lastWordFlag )
        ? flaggedChar
        : ( toCarry | (((*i) >> (4*gCodonBits + 2*gBaseBits))&maskBase ));

      *(pCodon++) = getCodonFromWord<3,2>(*i);
      *(pCodon++) = getCodonFromWord<2,2>(*i);
      *(pCodon++) = getCodonFromWord<1,2>(*i);
      *(pCodon++) = getCodonFromWord<0,2>(*i);

      lastWordFlag = (*i)&gCursedWord;
      toCarry = ((*i)& mask2Bases) << gBaseBits;
 
    } // ~for
    if (in.getNumBasesInLast() >= readingFrame )
    {
      *(pCodon++) = ( ((*i)&gCursedWord)|lastWordFlag )
        ? flaggedChar
        : ( toCarry | (((*i) >> (4*gCodonBits + 2*gBaseBits))&maskBase ));
    }
    if (codonsInLast>=1)
      *(pCodon++) = getCodonFromWord<3,2>(*i);
    if (codonsInLast>=2)
      *(pCodon++) = getCodonFromWord<2,2>(*i);
    if (codonsInLast>=3)
      *(pCodon++) = getCodonFromWord<1,2>(*i);
    if (codonsInLast>=4)
      *(pCodon++) = getCodonFromWord<0,2>(*i);

  } // ~else if
  else // if (readingFrame==2)
  {

    for( ; i!=lastWord ; ++i )
    {
      if (toCarry!=~0)
      (*pCodon++) = ( ((*i)&gCursedWord)|lastWordFlag )
        ? flaggedChar
        : ( toCarry | (((*i) >> (4*gCodonBits + gBaseBits))&mask2Bases ));

      *(pCodon++) = getCodonFromWord<3,1>(*i);
      *(pCodon++) = getCodonFromWord<2,1>(*i);
      *(pCodon++) = getCodonFromWord<1,1>(*i);
      *(pCodon++) = getCodonFromWord<0,1>(*i);

      lastWordFlag = (*i)&gCursedWord;
      toCarry = ((*i)& maskBase) << (2*gBaseBits);
 
    } // ~for
    if (in.getNumBasesInLast() >= readingFrame )
    {
    
      *(pCodon++) = ( ((*i)&gCursedWord)|lastWordFlag )
        ? flaggedChar
        : ( toCarry | (((*i) >> (4*gCodonBits + gBaseBits))&mask2Bases ));
    }
    if (codonsInLast>=1)
      *(pCodon++) = getCodonFromWord<3,1>(*i);
    if (codonsInLast>=2)
      *(pCodon++) = getCodonFromWord<2,1>(*i);
    if (codonsInLast>=3)
      *(pCodon++) = getCodonFromWord<1,1>(*i);
    if (codonsInLast>=4)
      *(pCodon++) = getCodonFromWord<0,1>(*i);

  } // ~else if

  //  cout << "codons fwd: " << codons << endl;
  assert(pCodon==pEnd);

} // ~void codonizeAndFlag

Here is the call graph for this function:

Here is the caller graph for this function:

void codonizeAndFlagReverse	(	const WordSequence &	in,
		CodonList &	codons,
		int	readingFrame
	)

Definition at line 536 of file SequenceEncoder.cpp.

References flaggedChar(), gBaseBits, gCodonBits, gCursedWord, WordSequence::getNumBasesInLast(), gMaxBasesPerWord, gNumReadingFrames, mask2Bases(), maskBase(), and maskCodon().

Referenced by HashTableTranslated::countWordsAndGetNames(), HashTableTranslated::hashAllWords(), and main().

{

  static const int DNAWordSizeForHashing(gMaxBasesPerWord-1);

  int codonsInLast((in.getNumBasesInLast()-readingFrame)/gNumReadingFrames);


  int numCodons
    (   ( (in.size() == 0) 
          ? 0
          : ((in.size()-1)*DNAWordSizeForHashing)+in.getNumBasesInLast()
          -readingFrame )
      / gNumReadingFrames );

  readingFrame = (in.getNumBasesInLast()-readingFrame+gNumReadingFrames)%gNumReadingFrames;



  //  cout << "blx: " << in.size() << " " << in.getNumBasesInLast() << " " << readingFrame << " " << codonsInLast << " " << numCodons << endl;
  codons.resize(numCodons);
  char* pCodon(static_cast<char*>(&(*codons.begin())));
  char* pEnd(static_cast<char*>(&(*codons.end())));
  WordSequence::const_iterator i(in.end()-1);
  WordSequence::const_iterator lastWord(in.begin());
  Word toCarry(~0);
  Word lastWordFlag(0);

  //  const Word maskBase(0x3);
  //  const Word maskCodon((0x3>>4)|(0x3>>2)|(0x3>>4));
  

  if (readingFrame==0)
  {
    if (codonsInLast==4)
      *(pCodon++) = getCodonFromWord<1,0>(*i);
    if (codonsInLast>=3)
      *(pCodon++) = getCodonFromWord<2,0>(*i);
    if (codonsInLast>=2)
      *(pCodon++) = getCodonFromWord<3,0>(*i);
    if (codonsInLast>=1)
      *(pCodon++) = getCodonFromWord<4,0>(*i);

    do
    {
      i--;
      *(pCodon++) = getCodonFromWord<0,0>(*i);
      *(pCodon++) = getCodonFromWord<1,0>(*i);
      *(pCodon++) = getCodonFromWord<2,0>(*i);
      *(pCodon++) = getCodonFromWord<3,0>(*i);
      *(pCodon++) = getCodonFromWord<4,0>(*i);

    } 
    while (i!=lastWord);

  } // ~if 
  else if (readingFrame==1)
  {
    //    cout << "rf1" << endl;
    if (codonsInLast==4)
      *(pCodon++) = getCodonFromWord<0,2>(*i);
    if (codonsInLast>=3)
      *(pCodon++) = getCodonFromWord<1,2>(*i);
    if (codonsInLast>=2)
      *(pCodon++) = getCodonFromWord<2,2>(*i);
    if (codonsInLast>=1)
      *(pCodon++) = getCodonFromWord<3,2>(*i);

    if (in.getNumBasesInLast()>=readingFrame)
    {
      lastWordFlag = (*i)&gCursedWord;
      toCarry = ((*i) >> (4*gCodonBits + 2*gBaseBits))&maskBase ;
    } // ~if

    do
    {
      i--;
      if (toCarry!=~0)
      (*pCodon++) = ( ((*i)&gCursedWord)|lastWordFlag )
        ? flaggedChar
        : ( toCarry | ( ((*i) & mask2Bases ) << gBaseBits ) );

      *(pCodon++) = getCodonFromWord<0,2>(*i);
      *(pCodon++) = getCodonFromWord<1,2>(*i);
      *(pCodon++) = getCodonFromWord<2,2>(*i);
      *(pCodon++) = getCodonFromWord<3,2>(*i);
      lastWordFlag = (*i)&gCursedWord;
      toCarry = ((*i) >> (4*gCodonBits + 2*gBaseBits))&maskBase ;


    } 
    while (i!=lastWord);


  } // ~else if
  else // if (readingFrame==2)
  {
    //    cout << "rf2" << endl;

    if (codonsInLast==4)
      *(pCodon++) = getCodonFromWord<0,1>(*i);
    if (codonsInLast>=3)
      *(pCodon++) = getCodonFromWord<1,1>(*i);
    if (codonsInLast>=2)
      *(pCodon++) = getCodonFromWord<2,1>(*i);
    if (codonsInLast>=1)
      *(pCodon++) = getCodonFromWord<3,1>(*i);

    if (in.getNumBasesInLast()>=readingFrame)
    {
      lastWordFlag = (*i)&gCursedWord;
      toCarry = ((*i) >> (4*gCodonBits + gBaseBits))&mask2Bases ;
    } // ~if

    do
    {
      i--;
      if (toCarry!=~0)
      (*pCodon++) = ( ((*i)&gCursedWord)|lastWordFlag )
        ? flaggedChar
        : ( toCarry | ( ((*i) & maskBase ) << (2*gBaseBits) ) );

      *(pCodon++) = getCodonFromWord<0,1>(*i);
      *(pCodon++) = getCodonFromWord<1,1>(*i);
      *(pCodon++) = getCodonFromWord<2,1>(*i);
      *(pCodon++) = getCodonFromWord<3,1>(*i);
      lastWordFlag = (*i)&gCursedWord;
      toCarry = ((*i) >> (4*gCodonBits + gBaseBits))&mask2Bases ;

    } 
    while (i!=lastWord);


  } // ~else if


  assert(pCodon==pEnd);
  //  cout << "codons unrev: " << codons << endl;

  for (CodonList::iterator j(codons.begin()); j!=codons.end(); ++j)
  {
    
    *j = 
      ( ((*j)==flaggedChar) 
        ? flaggedChar 
        :
        ( ((*j)&maskBase^maskBase) << (2*gBaseBits) )
        | ((*j)&(maskBase<<gBaseBits)^(maskBase<<gBaseBits))
        | ( ((*j)&(maskCodon^mask2Bases)^(maskCodon^mask2Bases)) 
            >> (2*gBaseBits) )  );
  } // ~for
  //  cout << "codons rev: " << codons << endl;

} // ~void codonizeAndFlagReverse

Here is the call graph for this function:

Here is the caller graph for this function:

static const Word firstCharInvalid

(

1<<

31

)

[static]

Referenced by SequenceEncoder::expandTranslationTable().

Here is the caller graph for this function:

static const char flaggedChar

(

'^'

)

[static]

Referenced by codonizeAndFlag(), codonizeAndFlagReverse(), getCodonFromWord(), and operator<<().

Here is the caller graph for this function:

template<int CODON_SHIFT, int BASE_SHIFT>

Word getCodonFromWord

(

const Word &

w

)

[inline]

Definition at line 526 of file SequenceEncoder.h.

References flaggedChar(), gBaseBits, gCodonBits, gCursedWord, and maskCodon().

{
  return 
  (
     (w&gCursedWord) 
     ? flaggedChar
     : (    ( w & ( maskCodon << ((CODON_SHIFT*gCodonBits) 
                                  + (BASE_SHIFT*gBaseBits)   ) ) )
         >> ((CODON_SHIFT*gCodonBits) + (BASE_SHIFT*gBaseBits) ) ) 
  );


} //~template <int CODON_SHIFT, int BASE_SHIFT > Word getCodonFromWord

Here is the call graph for this function:

static const Word mask2Bases

(

(maskBase<< gBaseBits)|

maskBase

)

[static]

Referenced by codonizeAndFlag(), and codonizeAndFlagReverse().

Here is the caller graph for this function:

static const Word maskBase

(

0x3

)

[static]

Referenced by codonizeAndFlag(), and codonizeAndFlagReverse().

Here is the caller graph for this function:

static const Word maskCodon

(

(mask2Bases<< gBaseBits)|

maskBase

)

[static]

Referenced by codonizeAndFlagReverse(), and getCodonFromWord().

Here is the caller graph for this function:

static const int numPossibleChars

(

1<<

8

)

[static]

Referenced by print().

Here is the caller graph for this function:

static const Word nv

(

0xFF

)

[static]

Referenced by MatchAlignerTranslated::codonize(), HashTableTranslated::countWordsAndGetNames(), SequenceEncoder::encodeChar(), SequenceEncoder::expandTranslationTable(), MatchAlignerTranslated::getCodon(), and processQuery().

Here is the caller graph for this function:

ostream& operator<<	(	ostream &	os,
		CodonList &	c
	)

Definition at line 693 of file SequenceEncoder.cpp.

References flaggedChar(), and gCodonNames.

{
  for (vector<char>::iterator i(c.begin()); i!=c.end(); i++ )
    os << (unsigned int)*i << "-" << 
      ((*i==flaggedChar) ? '!' : gCodonNames[(unsigned int)*i]) << " ";
  return os;
} // ~ostream& operator<<( ostream& os, CodonList& c )

Here is the call graph for this function:

void print

(

const TranslationTable &

tt

)

Definition at line 37 of file SequenceEncoder.cpp.

References numPossibleChars().

{
  for ( int i(0) ; i < numPossibleChars ; ++ i )
  {
    cout << i << ":";
    if (isgraph((uchar)i)) cout << " (" << (uchar)i << ") ";
    cout << "\t " << (int)tt[i];
    if (isgraph(tt[i])) cout << " (" << tt[i] << ") ";
    cout << endl;
  }
}

Here is the call graph for this function:

static const Word secondCharInvalid

(

1<<

30

)

[static]

Referenced by SequenceEncoder::expandTranslationTable().

Here is the caller graph for this function:

static const Word someCharInvalid

(

firstCharInvalid|

secondCharInvalid

)

[static]

Referenced by SequenceEncoder::encode().

Here is the caller graph for this function:

---

Variable Documentation

const TranslationTable ttCodon [static]

Definition at line 164 of file SequenceEncoder.h.

Referenced by MatchAlignerTranslated::codonize(), HashTableTranslated::countWordsAndGetNames(), and MatchAlignerTranslated::getCodon().

const TranslationTable ttCodonReverse [static]

Definition at line 246 of file SequenceEncoder.h.

const TranslationTable ttDNA [static]

Initial value:

 
{
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,  
00,nv,01,nv,nv,nv,02,nv,nv,nv,nv,nv,nv,nv,nv, 
nv,nv,nv,nv,03,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,  
00,nv,01,nv,nv,nv,02,nv,nv,nv,nv,nv,nv,nv,nv, 
nv,nv,nv,nv,03,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv
}

Definition at line 64 of file SequenceEncoder.h.

Referenced by MatchAlignerTranslated::codonize(), MatchAlignerTranslated::getCodon(), and processQuery().

const TranslationTable ttProtein [static]

Initial value:

 
{
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv, 0,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,  
 1, 3, 2, 3, 4, 5, 6, 7, 8,nv, 9,10,11,12,nv, 
13,14,15,16,17,20,18,19,20,21, 4,nv,nv,nv,nv,nv,
nv,  
 1, 3, 2, 3, 4, 5, 6, 7, 8,nv, 9,10,11,12,nv, 
13,14,15,16,17,20,18,19,20,21, 4,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,
nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv,nv
}

Definition at line 141 of file SequenceEncoder.h.

Referenced by HashTableTranslated::countWordsAndGetNames(), and processQuery().

---