void SearchUsingBacktracker::run()
{
Timer timer;
bool compressIntermediateBwts = true;
vector <BwtReaderBase *> inBwt( alphabetSize );
LetterCountEachPile countsPerPile, countsCumulative;
RangeStoreExternal r;
string bwtPrefix = searchParams_["input"];
string subset_ = "";
vector<string> kmerList;
if ( searchParams_["one kmer string"].isSet() )
{
kmerList.push_back( searchParams_["one kmer string"] );
}
else
{
assert( searchParams_["kmers input file"].isSet() );
string filename = searchParams_["kmers input file"];
ifstream ifs( filename );
string line;
// get 1st word of each line as kmer
while ( getline( ifs, line ) )
{
istringstream iss( line );
string kmer;
iss >> kmer;
assert( kmer.size() != 1 && "todo: 1-mer search" );
if ( kmer.size() >= 2 )
kmerList.push_back( kmer );
}
}
SequenceNumber originalIndex = 0;
for ( auto kmer : kmerList )
{
std::reverse( kmer.begin(), kmer.end() );
kmerList2.push_back( KmerSearchItem( kmer, 0, 0, originalIndex++ ) );
}
std::sort( kmerList2.begin(), kmerList2.end() );
Logger_if( LOG_SHOW_IF_VERY_VERBOSE )
{
Logger::out() << "kmerList2:" << endl;
for ( auto kmerItem : kmerList2 )
Logger::out() << " " << kmerItem.kmer << endl;
}
for ( int i( 0 ); i < alphabetSize; i++ )
{
stringstream fileNameSS;
fileNameSS << bwtPrefix << "-B0" << i;
string fileName = fileNameSS.str().c_str();
if ( compressIntermediateBwts == true )
{
if ( searchParams_["use indexing"].isSet() )
{
Logger_if( LOG_SHOW_IF_VERBOSE ) Logger::out() << "Using indexed BWT file" << endl;
inBwt[i] = new BwtReaderRunLengthIndex( fileName );
}
else
{
Logger_if( LOG_SHOW_IF_VERBOSE ) Logger::out() << "Using non-indexed BWT file" << endl;
inBwt[i] = new BwtReaderRunLength( fileName );
}
}
else
inBwt[i] = new BwtReaderASCII( fileName );
inBwt[i]->readAndCount( countsPerPile[i] );
}
countsCumulative = countsPerPile;
for ( int i( 1 ); i < alphabetSize; i++ )
countsCumulative[i] += countsCumulative[i - 1];
Logger_if( LOG_SHOW_IF_VERBOSE )
{
cout << "countsPerPile:" << endl;
countsPerPile.print();
cout << "countsCumulative:" << endl;
countsCumulative.print();
}
string currentWord_notUsed;
int startPosInKmerList2 = 0;
int endPosInKmerList2 = 0;
for ( int i( 1 ); i < alphabetSize; ++i )
{
for ( int j( 1 ); j < alphabetSize; ++j )
{
startPosInKmerList2 = endPosInKmerList2;
while ( endPosInKmerList2 < ( int )kmerList2.size() &&
kmerList2[endPosInKmerList2].kmer[0] == alphabet[i] &&
kmerList2[endPosInKmerList2].kmer[1] == alphabet[j] )
++endPosInKmerList2;
if ( countsPerPile[i].count_[j] != 0 )
{
if ( startPosInKmerList2 != endPosInKmerList2 )
r.addRange(
KmerSearchRange(
currentWord_notUsed,
countsCumulative[i - 1].count_[j],
countsPerPile[i].count_[j],
false
, startPosInKmerList2
, endPosInKmerList2
)
, j,
i,
subset_,
1
);
}
else
{
if ( startPosInKmerList2 != endPosInKmerList2 )
{
cout << "Oh nooo, I can't find: " << endl;
for ( int k = startPosInKmerList2; k < endPosInKmerList2; ++k )
{
string rev = kmerList2[k].kmer;
std::reverse( rev.begin(), rev.end() );
cout << " " << kmerList2[k].kmer << " (" << rev << ")" << endl;
}
}
}
} // ~for j
} // ~for i
LetterCount countsSoFar;
LetterNumber currentPos;
LetterNumber numRanges, numSingletonRanges;
r.clear();
for ( int cycle( 1 ); ; ++cycle )
{
Logger_if( LOG_LEVEL_NORMAL )
Logger::out() << "cycle: " << cycle << endl;
Logger_if( LOG_SHOW_IF_VERBOSE )
{
Logger::out() << " time now: " << timer.timeNow();
Logger::out() << " usage: " << timer << endl;
}
numRanges = 0;
numSingletonRanges = 0;
r.setCycleNum( cycle );
for ( int i( 1 ); i < alphabetSize; ++i )
{
inBwt[i]->rewindFile();
currentPos = 0;
countsSoFar.clear();
countsSoFar += countsCumulative[i - 1];
for ( int j( 1 ); j < alphabetSize; ++j )
{
r.setPortion( i, j );
OneBwtBackTracker backTracker(
inBwt[i],
currentPos,
r,
countsSoFar,
subset_,
cycle + 1,
false, // don't propagate breakpoints until $ sign
true // skip-already-processed-intervals deactivated
);
KmerSearchIntervalHandler intervalHandler;
KmerSearchRange rangeObject;
backTracker.process( i, currentWord_notUsed, intervalHandler, rangeObject );
numRanges += backTracker.numRanges_;
numSingletonRanges += backTracker.numSingletonRanges_;
} // ~for j
//cerr << "Done i " << i <<endl;
} // ~for i
r.clear();
// return 0; // %%%
Logger_if( LOG_SHOW_IF_VERBOSE )
{
Logger::out() << "Finished cycle " << cycle << ": ranges=" << numRanges << " singletons=" << numSingletonRanges << endl;
}
if ( numRanges == 0 ) break;
} // ~for c
for ( int i = 0; i < alphabetSize; i++ )
delete inBwt[i];
// Output
{
ostream *outputStreamPtr = &std::cout;
string outputFilename = searchParams_["output"];
ofstream ofs;
if ( outputFilename != "-" )
{
ofs.open( outputFilename );
if ( ofs.good() )
outputStreamPtr = &ofs;
else
cerr << "Warning: Couldn't open output file " << outputFilename << ". Sending output to stdout." << endl;
}
IntervalWriter writer( *outputStreamPtr );
for ( auto kmerItem : kmerList2 )
{
IntervalRecord rec( kmerList[kmerItem.originalIndex], kmerItem.position, kmerItem.count );
writer.write( rec );
}
}
}
ErrorStore BwtCorrector::findErrors()
{
const bool propagateSequence = false;//( compareParams_ ? ( *compareParams_ )["propagate sequence"] : false );
ErrorStore result;
Timer timer;
bool compressIntermediateBwts = true;
vector <BwtReaderBase *> inBwt( alphabetSize );
LetterCountEachPile countsPerPile, countsCumulative;
RangeStoreExternal r;
int numCycles( readLength_ );
// int minOcc( numberOfReads_ );
for ( int i( 0 ); i < alphabetSize; i++ )
{
stringstream fileNameSS;
fileNameSS << indexPrefix_ << "-B0" << i;
string fileName = fileNameSS.str().c_str();
if ( compressIntermediateBwts == true )
inBwt[i] = new BwtReaderRunLengthIndex( fileName, correctorParams_->getStringValue( "use shm" ) );
else
inBwt[i] = new BwtReaderASCII( fileName );
inBwt[i]-> readAndCount( countsPerPile[i] );
}
countsCumulative = countsPerPile;
for ( int i( 1 ); i < alphabetSize; i++ )
countsCumulative[i] += countsCumulative[i - 1];
countsCumulative.print();
string currentWord = "xx";
for ( int i( 1 ); i < alphabetSize; ++i )
{
if ( propagateSequence )
currentWord[1] = alphabet[i];
for ( int j( 1 ); j < alphabetSize; ++j )
{
if ( propagateSequence )
currentWord[0] = alphabet[j];
if ( countsPerPile[i].count_[j] != 0 )
r.addRange(
ErrorCorrectionRange(
currentWord,
countsCumulative[i - 1].count_[j],
countsPerPile[i].count_[j],
false
)
, j,
i,
subset_,
1
);
} // ~for j
} // ~for i
LetterCount countsSoFar;
LetterNumber currentPos;
LetterNumber numRanges, numSingletonRanges;
r.clear();
for ( int c( 0 ); c < numCycles; ++c )
{
int minimumSupport = getMinSupport( c );
cout << "cycle: " << c << endl;
Logger_if( LOG_SHOW_IF_VERBOSE )
{
Logger::out() << " time now: " << timer.timeNow();
Logger::out() << " usage: " << timer << endl;
}
numRanges = 0;
numSingletonRanges = 0;
r.setCycleNum( c + 1 );
for ( int i( 1 ); i < alphabetSize; ++i )
{
string thisWord( c + 3, 'x' );
inBwt[i]->rewindFile();
currentPos = 0;
countsSoFar.clear();
countsSoFar += countsCumulative[i - 1];
#ifdef PROB_NOT_NEEDED
BwtReaderRunLengthIndex *pRun;
pRun = dynamic_cast<BwtReaderRunLengthIndex *>( inBwt[i] );
if ( pRun != NULL )
pRun->initIndex( countsSoFar );
else
inBwtA[i]->rewindFile();
#endif
for ( int j( 1 ); j < alphabetSize; ++j )
{
r.setPortion( i, j );
OneBwtBackTracker backTracker(
inBwt[i],
currentPos,
r,
countsSoFar,
// numCycles,
subset_,
c + 2,
false, // doesn't propagate to end of reads
true, // skip-already-processed-intervals deactivated
propagateSequence,
endPosFile_
);
BwtCorrectorIntervalHandler intervalHandler( result, minWitnessLength_, minimumSupport, c + 2 );
ErrorCorrectionRange rangeObject;
backTracker.process( i, thisWord, intervalHandler, rangeObject );
numRanges += backTracker.numRanges_;
numSingletonRanges += backTracker.numSingletonRanges_;
} // ~for j
//cerr << "Done i " << i <<endl;
} // ~for i
r.clear();
// return 0; // %%%
Logger_if( LOG_SHOW_IF_VERBOSE )
{
Logger::out() << "Finished cycle " << c << ": ranges=" << numRanges << " singletons=" << numSingletonRanges << endl;
}
if ( numRanges == 0 ) break;
} // ~for c
for ( int i = 0; i < alphabetSize; i++ )
delete inBwt[i];
return result;
}