//
// Main
//
int correctMain(int argc, char** argv)
{
parseCorrectOptions(argc, argv);
BWT* pBWT = new BWT(opt::prefix + BWT_EXT, opt::sampleRate);
BWT* pRBWT = NULL;
// If the correction mode is k-mer only, then do not load the reverse
// BWT as it is not needed
if(opt::algorithm != ECA_KMER)
pRBWT = new BWT(opt::prefix + RBWT_EXT, opt::sampleRate);
BWTIntervalCache intervalCache(opt::intervalCacheLength, pBWT);
OverlapAlgorithm* pOverlapper = new OverlapAlgorithm(pBWT, NULL,
opt::errorRate, opt::seedLength,
opt::seedStride, false, opt::branchCutoff);
// Learn the parameters of the kmer corrector
if(opt::bLearnKmerParams)
{
int threshold = learnKmerParameters(pBWT);
if(threshold != -1)
CorrectionThresholds::Instance().setBaseMinSupport(threshold);
}
// Open outfiles and start a timer
std::ostream* pWriter = createWriter(opt::outFile);
std::ostream* pDiscardWriter = (!opt::discardFile.empty() ? createWriter(opt::discardFile) : NULL);
Timer* pTimer = new Timer(PROGRAM_IDENT);
pBWT->printInfo();
// Set the error correction parameters
ErrorCorrectParameters ecParams;
ecParams.pOverlapper = pOverlapper;
ecParams.pIntervalCache = &intervalCache;
ecParams.algorithm = opt::algorithm;
ecParams.minOverlap = opt::minOverlap;
ecParams.numOverlapRounds = opt::numOverlapRounds;
ecParams.conflictCutoff = opt::conflictCutoff;
ecParams.numKmerRounds = opt::numKmerRounds;
ecParams.kmerLength = opt::kmerLength;
ecParams.printOverlaps = opt::verbose > 1;
// Setup post-processor
bool bCollectMetrics = !opt::metricsFile.empty();
ErrorCorrectPostProcess postProcessor(pWriter, pDiscardWriter, bCollectMetrics);
if(opt::numThreads <= 1)
{
// Serial mode
ErrorCorrectProcess processor(ecParams);
SequenceProcessFramework::processSequencesSerial<SequenceWorkItem,
ErrorCorrectResult,
ErrorCorrectProcess,
ErrorCorrectPostProcess>(opt::readsFile, &processor, &postProcessor);
}
else
{
// Parallel mode
std::vector<ErrorCorrectProcess*> processorVector;
for(int i = 0; i < opt::numThreads; ++i)
{
ErrorCorrectProcess* pProcessor = new ErrorCorrectProcess(ecParams);
processorVector.push_back(pProcessor);
}
SequenceProcessFramework::processSequencesParallel<SequenceWorkItem,
ErrorCorrectResult,
ErrorCorrectProcess,
ErrorCorrectPostProcess>(opt::readsFile, processorVector, &postProcessor);
for(int i = 0; i < opt::numThreads; ++i)
{
delete processorVector[i];
}
}
if(bCollectMetrics)
{
std::ostream* pMetricsWriter = createWriter(opt::metricsFile);
postProcessor.writeMetrics(pMetricsWriter);
delete pMetricsWriter;
}
delete pBWT;
if(pRBWT != NULL)
delete pRBWT;
delete pOverlapper;
delete pTimer;
delete pWriter;
if(pDiscardWriter != NULL)
delete pDiscardWriter;
if(opt::numThreads > 1)
pthread_exit(NULL);
return 0;
}
//
// Main
//
int correctMain(int argc, char** argv)
{
parseCorrectOptions(argc, argv);
std::cout << "Correcting sequencing errors for " << opt::readsFile << "\n";
// Load indices
BWT* pBWT = new BWT(opt::prefix + BWT_EXT, opt::sampleRate);
BWT* pRBWT = NULL;
SampledSuffixArray* pSSA = NULL;
if(opt::algorithm == ECA_OVERLAP)
pSSA = new SampledSuffixArray(opt::prefix + SAI_EXT, SSA_FT_SAI);
BWTIntervalCache* pIntervalCache = new BWTIntervalCache(opt::intervalCacheLength, pBWT);
BWTIndexSet indexSet;
indexSet.pBWT = pBWT;
indexSet.pRBWT = pRBWT;
indexSet.pSSA = pSSA;
indexSet.pCache = pIntervalCache;
// Learn the parameters of the kmer corrector
if(opt::bLearnKmerParams)
{
int threshold = learnKmerParameters(pBWT);
if(threshold != -1)
CorrectionThresholds::Instance().setBaseMinSupport(threshold);
}
// Open outfiles and start a timer
std::ostream* pWriter = createWriter(opt::outFile);
std::ostream* pDiscardWriter = (!opt::discardFile.empty() ? createWriter(opt::discardFile) : NULL);
Timer* pTimer = new Timer(PROGRAM_IDENT);
pBWT->printInfo();
// Set the error correction parameters
ErrorCorrectParameters ecParams;
ecParams.pOverlapper = NULL;
ecParams.indices = indexSet;
ecParams.algorithm = opt::algorithm;
ecParams.minOverlap = opt::minOverlap;
ecParams.numOverlapRounds = opt::numOverlapRounds;
ecParams.minIdentity = 1.0f - opt::errorRate;
ecParams.conflictCutoff = opt::conflictCutoff;
ecParams.numKmerRounds = opt::numKmerRounds;
ecParams.kmerLength = opt::kmerLength;
ecParams.printOverlaps = opt::verbose > 0;
// Setup post-processor
bool bCollectMetrics = !opt::metricsFile.empty();
ErrorCorrectPostProcess postProcessor(pWriter, pDiscardWriter, bCollectMetrics);
if(opt::numThreads <= 1)
{
// Serial mode
ErrorCorrectProcess processor(ecParams);
SequenceProcessFramework::processSequencesSerial<SequenceWorkItem,
ErrorCorrectResult,
ErrorCorrectProcess,
ErrorCorrectPostProcess>(opt::readsFile, &processor, &postProcessor);
}
else
{
// Parallel mode
std::vector<ErrorCorrectProcess*> processorVector;
for(int i = 0; i < opt::numThreads; ++i)
{
ErrorCorrectProcess* pProcessor = new ErrorCorrectProcess(ecParams);
processorVector.push_back(pProcessor);
}
SequenceProcessFramework::processSequencesParallel<SequenceWorkItem,
ErrorCorrectResult,
ErrorCorrectProcess,
ErrorCorrectPostProcess>(opt::readsFile, processorVector, &postProcessor);
for(int i = 0; i < opt::numThreads; ++i)
{
delete processorVector[i];
}
}
if(bCollectMetrics)
{
std::ostream* pMetricsWriter = createWriter(opt::metricsFile);
postProcessor.writeMetrics(pMetricsWriter);
delete pMetricsWriter;
}
delete pBWT;
delete pIntervalCache;
if(pRBWT != NULL)
delete pRBWT;
if(pSSA != NULL)
delete pSSA;
delete pTimer;
delete pWriter;
if(pDiscardWriter != NULL)
delete pDiscardWriter;
if(opt::numThreads > 1)
pthread_exit(NULL);
return 0;
}
