void rmdup()
{
StringVector hitsFilenames;
BWT* pBWT = new BWT(opt::prefix + BWT_EXT, opt::sampleRate);
BWT* pRBWT = new BWT(opt::prefix + RBWT_EXT, opt::sampleRate);
OverlapAlgorithm* pOverlapper = new OverlapAlgorithm(pBWT, pRBWT,
opt::errorRate, 0,
0, false);
Timer* pTimer = new Timer(PROGRAM_IDENT);
if(opt::numThreads <= 1)
{
printf("[%s] starting serial-mode overlap computation\n", PROGRAM_IDENT);
computeRmdupHitsSerial(opt::prefix, opt::readsFile, pOverlapper, hitsFilenames);
}
else
{
printf("[%s] starting parallel-mode overlap computation with %d threads\n", PROGRAM_IDENT, opt::numThreads);
computeRmdupHitsParallel(opt::numThreads, opt::prefix, opt::readsFile, pOverlapper, hitsFilenames);
}
delete pOverlapper;
delete pBWT;
delete pRBWT;
delete pTimer;
std::string out_prefix = stripExtension(opt::outFile);
std::string dupsFile = parseDupHits(hitsFilenames, out_prefix);
// Rebuild the indices without the duplicated sequences
if(opt::bReindex)
{
std::cout << "Rebuilding indices without duplicated reads\n";
removeReadsFromIndices(opt::prefix, dupsFile, out_prefix, BWT_EXT, SAI_EXT, false, opt::numThreads);
removeReadsFromIndices(opt::prefix, dupsFile, out_prefix, RBWT_EXT, RSAI_EXT, true, opt::numThreads);
}
}
//
// Main
//
int filterMain(int argc, char** argv)
{
parseFilterOptions(argc, argv);
Timer* pTimer = new Timer(PROGRAM_IDENT);
BWT* pBWT = new BWT(opt::prefix + BWT_EXT, opt::sampleRate);
BWT* pRBWT = new BWT(opt::prefix + RBWT_EXT, opt::sampleRate);
pBWT->printInfo();
std::ostream* pWriter = createWriter(opt::outFile);
std::ostream* pDiscardWriter = createWriter(opt::discardFile);
QCPostProcess* pPostProcessor = new QCPostProcess(pWriter, pDiscardWriter);
// If performing duplicate check, create a bitvector to record
// which reads are duplicates
BitVector* pSharedBV = NULL;
if(opt::dupCheck)
pSharedBV = new BitVector(pBWT->getNumStrings());
// Set up QC parameters
QCParameters params;
params.pBWT = pBWT;
params.pRevBWT = pRBWT;
params.pSharedBV = pSharedBV;
params.checkDuplicates = opt::dupCheck;
params.substringOnly = opt::substringOnly;
params.checkKmer = opt::kmerCheck;
params.kmerBothStrand = opt::kmerBothStrand;
params.checkHPRuns = opt::hpCheck;
params.checkDegenerate = opt::lowComplexityCheck;
params.verbose = opt::verbose;
params.kmerLength = opt::kmerLength;
params.kmerThreshold = opt::kmerThreshold;
params.hpKmerLength = 51;
params.hpHardAcceptCount = 10;
params.hpMinProportion = 0.1f;
params.hpMinLength = 6;
if(opt::numThreads <= 1)
{
// Serial mode
QCProcess processor(params);
PROCESS_FILTER_SERIAL(opt::readsFile, &processor, pPostProcessor);
}
else
{
// Parallel mode
std::vector<QCProcess*> processorVector;
for(int i = 0; i < opt::numThreads; ++i)
{
QCProcess* pProcessor = new QCProcess(params);
processorVector.push_back(pProcessor);
}
PROCESS_FILTER_PARALLEL(opt::readsFile, processorVector, pPostProcessor);
for(int i = 0; i < opt::numThreads; ++i)
delete processorVector[i];
}
delete pPostProcessor;
delete pWriter;
delete pDiscardWriter;
delete pBWT;
delete pRBWT;
if(pSharedBV != NULL)
delete pSharedBV;
// Rebuild the FM-index without the discarded reads
std::string out_prefix = stripFilename(opt::outFile);
removeReadsFromIndices(opt::prefix, opt::discardFile, out_prefix, BWT_EXT, SAI_EXT, false, opt::numThreads);
removeReadsFromIndices(opt::prefix, opt::discardFile, out_prefix, RBWT_EXT, RSAI_EXT, true, opt::numThreads);
// Cleanup
delete pTimer;
if(opt::numThreads > 1)
pthread_exit(NULL);
return 0;
}
