Esempio n. 1
0
int32_t elog::registerInstance(const std::string& _name) {
	for (size_t iii = 0; iii < getList().size(); ++iii) {
		if (getList()[iii].first == _name) {
			return iii;
		}
	}
	getList().push_back(std::make_pair(_name, getDefaultLevel()));
	if (_name.size() >= getNameSizeLog()) {
		getNameSizeLog() = _name.size()+1;
	}
	//std::cout << "register log : '" << _name << "'=" << getList().size()-1 << std::endl;
	return getList().size()-1;
}
Esempio n. 2
0
int main(int argc, char * argv[])
{
	try
	{
		::libmaus::util::ArgInfo const arginfo(argc,argv);

		for ( uint64_t i = 0; i < arginfo.restargs.size(); ++i )
			if ( 
				arginfo.restargs[i] == "-v"
				||
				arginfo.restargs[i] == "--version"
			)
			{
				std::cerr << ::biobambam::Licensing::license();
				return EXIT_SUCCESS;
			}
			else if ( 
				arginfo.restargs[i] == "-h"
				||
				arginfo.restargs[i] == "--help"
			)
			{
				std::cerr << ::biobambam::Licensing::license();
				std::cerr << std::endl;
				std::cerr << "Key=Value pairs:" << std::endl;
				std::cerr << std::endl;
				
				std::vector< std::pair<std::string,std::string> > V;
				
				V.push_back ( std::pair<std::string,std::string> ( "verbose=<["+::biobambam::Licensing::formatNumber(getDefaultVerbose())+"]>", "print progress report" ) );
				V.push_back ( std::pair<std::string,std::string> ( "colhashbits=<["+::biobambam::Licensing::formatNumber(getDefaultColHashBits())+"]>", "log_2 of size of hash table used for collation" ) );
				V.push_back ( std::pair<std::string,std::string> ( "collistsize=<["+::biobambam::Licensing::formatNumber(getDefaultColListSize())+"]>", "output list size for collation" ) );
				V.push_back ( std::pair<std::string,std::string> ( "tmpfile=<filename>", "prefix for temporary files, default: create files in current directory" ) );
				V.push_back ( std::pair<std::string,std::string> ( "level=<["+::biobambam::Licensing::formatNumber(getDefaultLevel())+"]>", "compression settings for output bam file (0=uncompressed,1=fast,9=best,-1=zlib default)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5=<["+::biobambam::Licensing::formatNumber(getDefaultMD5())+"]>", "create md5 check sum (default: 0)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5filename=<filename>", "file name for md5 check sum (default: extend output file name)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "index=<["+::biobambam::Licensing::formatNumber(getDefaultIndex())+"]>", "create BAM index (default: 0)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "indexfilename=<filename>", "file name for BAM index file (default: extend output file name)" ) );

				::biobambam::Licensing::printMap(std::cerr,V);

				std::cerr << std::endl;
				return EXIT_SUCCESS;
			}
		
		return bamfixmatecoordinates(arginfo);
	}
	catch(std::exception const & ex)
	{
		std::cerr << ex.what() << std::endl;
		return EXIT_FAILURE;
	}
}
Esempio n. 3
0
int main(int argc, char * argv[])
{
	try
	{
		::libmaus::util::ArgInfo const arginfo(argc,argv);
		
		for ( uint64_t i = 0; i < arginfo.restargs.size(); ++i )
			if ( 
				arginfo.restargs[i] == "-v"
				||
				arginfo.restargs[i] == "--version"
			)
			{
				std::cerr << ::biobambam::Licensing::license();
				return EXIT_SUCCESS;
			}
			else if ( 
				arginfo.restargs[i] == "-h"
				||
				arginfo.restargs[i] == "--help"
			)
			{
				std::cerr << ::biobambam::Licensing::license();
				std::cerr << std::endl;
				std::cerr << "Key=Value pairs:" << std::endl;
				std::cerr << std::endl;
				
				std::vector< std::pair<std::string,std::string> > V;
			
				V.push_back ( std::pair<std::string,std::string> ( "level=<["+::biobambam::Licensing::formatNumber(getDefaultLevel())+"]>", "compression settings for output bam file (0=uncompressed,1=fast,9=best,-1=zlib default)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "SO=<["+getDefaultSortOrder()+"]>", "sorting order (coordinate or queryname)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "verbose=<["+::biobambam::Licensing::formatNumber(getDefaultVerbose())+"]>", "print progress report" ) );
				V.push_back ( std::pair<std::string,std::string> ( "blockmb=<["+::biobambam::Licensing::formatNumber(getDefaultBlockSize())+"]>", "size of internal memory buffer used for sorting in MiB" ) );
				V.push_back ( std::pair<std::string,std::string> ( "disablevalidation=<["+::biobambam::Licensing::formatNumber(getDefaultDisableValidation())+"]>", "disable input validation (default is 0)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "tmpfile=<filename>", "prefix for temporary files, default: create files in current directory" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5=<["+::biobambam::Licensing::formatNumber(getDefaultMD5())+"]>", "create md5 check sum (default: 0)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5filename=<filename>", "file name for md5 check sum" ) );
				V.push_back ( std::pair<std::string,std::string> ( "index=<["+::biobambam::Licensing::formatNumber(getDefaultIndex())+"]>", "create BAM index (default: 0)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "indexfilename=<filename>", "file name for BAM index file" ) );
				V.push_back ( std::pair<std::string,std::string> ( std::string("inputformat=<[")+getDefaultInputFormat()+"]>", std::string("input format (") + libmaus::bambam::BamMultiAlignmentDecoderFactory::getValidInputFormats() + ")" ) );
				V.push_back ( std::pair<std::string,std::string> ( std::string("outputformat=<[")+libmaus::bambam::BamBlockWriterBaseFactory::getDefaultOutputFormat()+"]>", std::string("output format (") + libmaus::bambam::BamBlockWriterBaseFactory::getValidOutputFormats() + ")" ) );
				V.push_back ( std::pair<std::string,std::string> ( "I=<[stdin]>", "input filename (standard input if unset)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "inputthreads=<[1]>", "input helper threads (for inputformat=bam only, default: 1)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "reference=<>", "reference FastA (.fai file required, for cram i/o only)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "range=<>", "coordinate range to be processed (for coordinate sorted indexed BAM input only)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "outputthreads=<[1]>", "output helper threads (for outputformat=bam only, default: 1)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "O=<[stdout]>", "output filename (standard output if unset)" ) );
				V.push_back ( std::pair<std::string,std::string> ( std::string("fixmates=<[")+::biobambam::Licensing::formatNumber(getDefaultFixMates())+"]>", "fix mate information (for name collated input only, disabled by default)" ) );

				::biobambam::Licensing::printMap(std::cerr,V);

				std::cerr << std::endl;
				return EXIT_SUCCESS;
			}
			
		return bamsort(arginfo);
	}
	catch(std::exception const & ex)
	{
		std::cerr << ex.what() << std::endl;
		return EXIT_FAILURE;
	}
}
Esempio n. 4
0
int main(int argc, char * argv[])
{
	try
	{
		::libmaus2::util::ArgInfo const arginfo(argc,argv);

		for ( uint64_t i = 0; i < arginfo.restargs.size(); ++i )
			if (
				arginfo.restargs[i] == "-v"
				||
				arginfo.restargs[i] == "--version"
			)
			{
				std::cerr << ::biobambam2::Licensing::license();
				return EXIT_SUCCESS;
			}
			else if (
				arginfo.restargs[i] == "-h"
				||
				arginfo.restargs[i] == "--help"
			)
			{
				std::cerr << ::biobambam2::Licensing::license();
				std::cerr << std::endl;
				std::cerr << "Key=Value pairs:" << std::endl;
				std::cerr << std::endl;

				std::vector< std::pair<std::string,std::string> > V;

				V.push_back ( std::pair<std::string,std::string> ( "level=<["+::biobambam2::Licensing::formatNumber(getDefaultLevel())+"]>", libmaus2::bambam::BamBlockWriterBaseFactory::getBamOutputLevelHelpText() ) );
				V.push_back ( std::pair<std::string,std::string> ( "verbose=<["+::biobambam2::Licensing::formatNumber(getDefaultVerbose())+"]>", "print progress information" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5=<["+::biobambam2::Licensing::formatNumber(getDefaultMD5())+"]>", "create md5 check sum (default: 0)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5filename=<filename>", "file name for md5 check sum (default: extend output file name)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "index=<["+::biobambam2::Licensing::formatNumber(getDefaultIndex())+"]>", "create BAM index (default: 0)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "indexfilename=<filename>", "file name for BAM index file (default: extend output file name)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "tmpfile=<filename>", "prefix for temporary files, default: create files in current directory" ) );

				::biobambam2::Licensing::printMap(std::cerr,V);

				std::cerr << std::endl;

				return EXIT_SUCCESS;
			}

		return bamclipreinsert(arginfo);
	}
	catch(std::exception const & ex)
	{
		std::cerr << ex.what() << std::endl;
		return EXIT_FAILURE;
	}
}
Esempio n. 5
0
int main(int argc, char * argv[])
{
	try
	{
		libmaus::timing::RealTimeClock rtc; rtc.start();
		
		::libmaus::util::ArgInfo const arginfo(argc,argv);
		
		for ( uint64_t i = 0; i < arginfo.restargs.size(); ++i )
			if ( 
				arginfo.restargs[i] == "-v"
				||
				arginfo.restargs[i] == "--version"
			)
			{
				std::cerr << ::biobambam::Licensing::license();
				return EXIT_SUCCESS;
			}
			else if ( 
				arginfo.restargs[i] == "-h"
				||
				arginfo.restargs[i] == "--help"
			)
			{
				std::cerr << ::biobambam::Licensing::license() << std::endl;
				std::cerr << "Key=Value pairs:" << std::endl;
				std::cerr << std::endl;
				
				std::vector< std::pair<std::string,std::string> > V;

				V.push_back ( std::pair<std::string,std::string> ( std::string("cols=<[")+libmaus::util::NumberSerialisation::formatNumber(getDefaultCols(),0)+"]>", "column width" ) );
				V.push_back ( std::pair<std::string,std::string> ( std::string("bgzf=<[")+libmaus::util::NumberSerialisation::formatNumber(getDefaultBgzf(),0)+"]>", "compress output" ) );
				V.push_back ( std::pair<std::string,std::string> ( std::string("index=<>"), "file name for index if bgzf=1 (no index is created if key is not given)" ) );
				V.push_back ( std::pair<std::string,std::string> ( std::string("level=<[")+::biobambam::Licensing::formatNumber(getDefaultLevel())+"]>", std::string("compression level if bgzf=1 (") + libmaus::bambam::BamBlockWriterBaseFactory::getLevelHelpText() + std::string(")") ) );
				
				::biobambam::Licensing::printMap(std::cerr,V);

				std::cerr << std::endl;
				return EXIT_SUCCESS;
			}
		
		normalisefasta(arginfo);
	}
	catch(std::exception const & ex)
	{
		std::cerr << ex.what() << std::endl;
		return EXIT_FAILURE;
	}

}
Esempio n. 6
0
int main(int argc, char * argv[])
{
	try
	{
		libmaus::util::ArgInfo const arginfo(argc,argv);
		
		for ( uint64_t i = 0; i < arginfo.restargs.size(); ++i )
			if ( 
				arginfo.restargs[i] == "-v"
				||
				arginfo.restargs[i] == "--version"
			)
			{
				std::cerr << ::biobambam::Licensing::license();
				return EXIT_SUCCESS;
			}
			else if ( 
				arginfo.restargs[i] == "-h"
				||
				arginfo.restargs[i] == "--help"
			)
			{
				std::cerr << ::biobambam::Licensing::license();
				std::cerr << std::endl;
				std::cerr << "Key=Value pairs:" << std::endl;
				std::cerr << std::endl;
				
				std::vector< std::pair<std::string,std::string> > V;
			
				V.push_back ( std::pair<std::string,std::string> ( "level=<["+::biobambam::Licensing::formatNumber(getDefaultLevel())+"]>", libmaus::bambam::BamBlockWriterBaseFactory::getBamOutputLevelHelpText() ) );
				V.push_back ( std::pair<std::string,std::string> ( "verbose=<["+::biobambam::Licensing::formatNumber(getDefaultVerbose())+"]>", "print progress report" ) );
				V.push_back ( std::pair<std::string,std::string> ( "I=<[input filename]>", "name of the input file (mandatory)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5=<["+::biobambam::Licensing::formatNumber(getDefaultMD5())+"]>", "create md5 check sum (default: 0)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5filename=<filename>", "file name for md5 check sum (default: extend output file name)" ) );

				::biobambam::Licensing::printMap(std::cerr,V);

				std::cerr << std::endl;
				return EXIT_SUCCESS;
			}

		return bamfilterheader2(arginfo);		
	}
	catch(std::exception const & ex)
	{
		std::cerr << ex.what() << std::endl;
		return EXIT_FAILURE;
	}
}
Esempio n. 7
0
int main(int argc, char * argv[])
{
	try
	{
		::libmaus::util::ArgInfo const arginfo(argc,argv);
		
		for ( uint64_t i = 0; i < arginfo.restargs.size(); ++i )
			if ( 
				arginfo.restargs[i] == "-v"
				||
				arginfo.restargs[i] == "--version"
			)
			{
				std::cerr << ::biobambam::Licensing::license();
				return EXIT_SUCCESS;
			}
			else if ( 
				arginfo.restargs[i] == "-h"
				||
				arginfo.restargs[i] == "--help"
			)
			{
				std::cerr << ::biobambam::Licensing::license();
				std::cerr << std::endl;
				std::cerr << "Key=Value pairs:" << std::endl;
				std::cerr << std::endl;
				
				std::vector< std::pair<std::string,std::string> > V;
				
				V.push_back ( std::pair<std::string,std::string> ( "level=<["+::biobambam::Licensing::formatNumber(getDefaultLevel())+"]>", libmaus::bambam::BamBlockWriterBaseFactory::getBamOutputLevelHelpText() ) );
				V.push_back ( std::pair<std::string,std::string> ( "minmapped=<["+::biobambam::Licensing::formatNumber(getDefaultMinMapped())+"]>", "minimum number of mapped fragments in a template" ) );
				V.push_back ( std::pair<std::string,std::string> ( "maxmapped=<["+::biobambam::Licensing::formatNumber(getDefaultMaxMapped())+"]>", "maximum number of mapped fragments in a template" ) );
				V.push_back ( std::pair<std::string,std::string> ( "minlen=<["+::biobambam::Licensing::formatNumber(getDefaultMinLen())+"]>", "minimum sequence length" ) );
				V.push_back ( std::pair<std::string,std::string> ( "verbose=<["+::biobambam::Licensing::formatNumber(getDefaultVerbose())+"]>", "print progress report (default: 1)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5=<["+::biobambam::Licensing::formatNumber(getDefaultMD5())+"]>", "create md5 check sum (default: 0)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5filename=<filename>", "file name for md5 check sum (default: extend output file name)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "index=<["+::biobambam::Licensing::formatNumber(getDefaultIndex())+"]>", "create BAM index (default: 0)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "indexfilename=<filename>", "file name for BAM index file (default: extend output file name)" ) );

				::biobambam::Licensing::printMap(std::cerr,V);

				std::cerr << std::endl;
				return EXIT_SUCCESS;
			}

		return bamfilter(arginfo);
	}
	catch(std::exception const & ex)
	{
		std::cerr << ex.what() << std::endl;
		return EXIT_FAILURE;
	}
}
Esempio n. 8
0
int bamreset(::libmaus2::util::ArgInfo const & arginfo)
{
	if ( isatty(STDIN_FILENO) )
	{
		::libmaus2::exception::LibMausException se;
		se.getStream() << "Refusing to read binary data from terminal, please redirect standard input to pipe or file." << std::endl;
		se.finish();
		throw se;
	}

	if ( isatty(STDOUT_FILENO) )
	{
		::libmaus2::exception::LibMausException se;
		se.getStream() << "Refusing write binary data to terminal, please redirect standard output to pipe or file." << std::endl;
		se.finish();
		throw se;
	}
	
	int const level = libmaus2::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(arginfo.getValue<int>("level",getDefaultLevel()));
	int const verbose = arginfo.getValue<int>("verbose",getDefaultVerbose());
	int const resetsortorder = arginfo.getValue<int>("resetsortorder",getDefaultResetSortOrder());
	
	::libmaus2::bambam::BamDecoder dec(std::cin,false);
	::libmaus2::bambam::BamHeader const & header = dec.getHeader();

	std::string headertext = header.text;

	// no replacement header file given
	if ( ! arginfo.hasArg("resetheadertext") )
	{
		// remove SQ lines
		std::vector<libmaus2::bambam::HeaderLine> allheaderlines = libmaus2::bambam::HeaderLine::extractLines(headertext);

		std::ostringstream upheadstr;
		for ( uint64_t i = 0; i < allheaderlines.size(); ++i )
			if ( allheaderlines[i].type != "SQ" )
				upheadstr << allheaderlines[i].line << std::endl;

		headertext = upheadstr.str();
	}
	// replace header given in file
	else
	{
		std::string const headerfilename = arginfo.getUnparsedValue("resetheadertext","");
		uint64_t const headerlen = libmaus2::util::GetFileSize::getFileSize(headerfilename);
		libmaus2::aio::CheckedInputStream CIS(headerfilename);
		libmaus2::autoarray::AutoArray<char> ctext(headerlen,false);
		CIS.read(ctext.begin(),headerlen);
		headertext = std::string(ctext.begin(),ctext.end());		
	}

	// add PG line to header
	headertext = libmaus2::bambam::ProgramHeaderLineSet::addProgramLine(
		headertext,
		"bamreset", // ID
		"bamreset", // PN
		arginfo.commandline, // CL
		::libmaus2::bambam::ProgramHeaderLineSet(headertext).getLastIdInChain(), // PP
		std::string(PACKAGE_VERSION) // VN			
	);
	
	// construct new header
	libmaus2::bambam::BamHeader uphead(headertext);
	if ( resetsortorder )
		uphead.changeSortOrder("unknown");

	/*
	 * start index/md5 callbacks
	 */
	std::string const tmpfilenamebase = arginfo.getValue<std::string>("tmpfile",arginfo.getDefaultTmpFileName());
	std::string const tmpfileindex = tmpfilenamebase + "_index";
	::libmaus2::util::TempFileRemovalContainer::addTempFile(tmpfileindex);
	uint32_t const excludeflags = libmaus2::bambam::BamFlagBase::stringToFlags(
		arginfo.getValue<std::string>("exclude",getDefaultExcludeFlags()));

	std::string md5filename;
	std::string indexfilename;

	std::vector< ::libmaus2::lz::BgzfDeflateOutputCallback * > cbs;
	::libmaus2::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Pmd5cb;
	if ( arginfo.getValue<unsigned int>("md5",getDefaultMD5()) )
	{
		if ( arginfo.hasArg("md5filename") &&  arginfo.getUnparsedValue("md5filename","") != "" )
			md5filename = arginfo.getUnparsedValue("md5filename","");
		else
			std::cerr << "[V] no filename for md5 given, not creating hash" << std::endl;

		if ( md5filename.size() )
		{
			::libmaus2::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Tmd5cb(new ::libmaus2::lz::BgzfDeflateOutputCallbackMD5);
			Pmd5cb = UNIQUE_PTR_MOVE(Tmd5cb);
			cbs.push_back(Pmd5cb.get());
		}
	}
	libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Pindex;
	if ( arginfo.getValue<unsigned int>("index",getDefaultIndex()) )
	{
		if ( arginfo.hasArg("indexfilename") &&  arginfo.getUnparsedValue("indexfilename","") != "" )
			indexfilename = arginfo.getUnparsedValue("indexfilename","");
		else
			std::cerr << "[V] no filename for index given, not creating index" << std::endl;

		if ( indexfilename.size() )
		{
			libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Tindex(new libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex(tmpfileindex));
			Pindex = UNIQUE_PTR_MOVE(Tindex);
			cbs.push_back(Pindex.get());
		}
	}
	std::vector< ::libmaus2::lz::BgzfDeflateOutputCallback * > * Pcbs = 0;
	if ( cbs.size() )
		Pcbs = &cbs;
	/*
	 * end md5/index callbacks
	 */

	::libmaus2::bambam::BamWriter::unique_ptr_type writer(new ::libmaus2::bambam::BamWriter(std::cout,uphead,level,Pcbs));
 	libmaus2::timing::RealTimeClock rtc; rtc.start();
 	
	libmaus2::bambam::BamAlignment & algn = dec.getAlignment();
	uint64_t c = 0;

	bool const resetaux = arginfo.getValue<int>("resetaux",getDefaultResetAux());
	libmaus2::bambam::BamAuxFilterVector::unique_ptr_type const prgfilter(libmaus2::bambam::BamAuxFilterVector::parseAuxFilterList(arginfo));
	libmaus2::bambam::BamAuxFilterVector const * rgfilter = prgfilter.get();

	while ( dec.readAlignment() )
	{
		bool const keep = resetAlignment(algn,resetaux /* reset aux */,excludeflags,rgfilter);
		
		if ( keep )
			algn.serialise(writer->getStream());

		if ( verbose && (++c & (1024*1024-1)) == 0 )
 			std::cerr << "[V] " << c/(1024*1024) << " " << (c / rtc.getElapsedSeconds()) << std::endl;
	}
	
	writer.reset();

	if ( Pmd5cb )
	{
		Pmd5cb->saveDigestAsFile(md5filename);
	}
	if ( Pindex )
	{
		Pindex->flush(std::string(indexfilename));
	}

	return EXIT_SUCCESS;
}
Esempio n. 9
0
int bam12split(::libmaus::util::ArgInfo const & arginfo)
{
	::libmaus::util::TempFileRemovalContainer::setup();

	if ( isatty(STDIN_FILENO) )
	{
		::libmaus::exception::LibMausException se;
		se.getStream() << "Refusing to read binary data from terminal, please redirect standard input to pipe or file." << std::endl;
		se.finish();
		throw se;
	}

	if ( isatty(STDOUT_FILENO) )
	{
		::libmaus::exception::LibMausException se;
		se.getStream() << "Refusing write binary data to terminal, please redirect standard output to pipe or file." << std::endl;
		se.finish();
		throw se;
	}
	
	if ( arginfo.hasArg("keep") && arginfo.hasArg("remove") )
	{
		::libmaus::exception::LibMausException se;
		se.getStream() << "The keep and remove keys are mutually exclusive." << std::endl;
		se.finish();
		throw se;		
	}

	int const level = libmaus::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(arginfo.getValue<int>("level",getDefaultLevel()));
	int const verbose = arginfo.getValue<int>("verbose",getDefaultVerbose());

	::libmaus::bambam::BamDecoder dec(std::cin,false);
	::libmaus::bambam::BamHeader const & header = dec.getHeader();

	std::string const headertext(header.text);

	// add PG line to header
	std::string const upheadtext = ::libmaus::bambam::ProgramHeaderLineSet::addProgramLine(
		headertext,
		"bam12split", // ID
		"bam12split", // PN
		arginfo.commandline, // CL
		::libmaus::bambam::ProgramHeaderLineSet(headertext).getLastIdInChain(), // PP
		std::string(PACKAGE_VERSION) // VN			
	);
	// construct new header
	libmaus::bambam::BamHeader uphead(upheadtext);
	uphead.changeSortOrder("unknown");

	/*
	 * start index/md5 callbacks
	 */
	std::string const tmpfilenamebase = arginfo.getValue<std::string>("tmpfile",arginfo.getDefaultTmpFileName());
	std::string const tmpfileindex = tmpfilenamebase + "_index";
	::libmaus::util::TempFileRemovalContainer::addTempFile(tmpfileindex);

	std::string md5filename;
	std::string indexfilename;

	std::vector< ::libmaus::lz::BgzfDeflateOutputCallback * > cbs;
	::libmaus::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Pmd5cb;
	if ( arginfo.getValue<unsigned int>("md5",getDefaultMD5()) )
	{
		if ( arginfo.hasArg("md5filename") &&  arginfo.getUnparsedValue("md5filename","") != "" )
			md5filename = arginfo.getUnparsedValue("md5filename","");
		else
			std::cerr << "[V] no filename for md5 given, not creating hash" << std::endl;

		if ( md5filename.size() )
		{
			::libmaus::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Tmd5cb(new ::libmaus::lz::BgzfDeflateOutputCallbackMD5);
			Pmd5cb = UNIQUE_PTR_MOVE(Tmd5cb);
			cbs.push_back(Pmd5cb.get());
		}
	}
	libmaus::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Pindex;
	if ( arginfo.getValue<unsigned int>("index",getDefaultIndex()) )
	{
		if ( arginfo.hasArg("indexfilename") &&  arginfo.getUnparsedValue("indexfilename","") != "" )
			indexfilename = arginfo.getUnparsedValue("indexfilename","");
		else
			std::cerr << "[V] no filename for index given, not creating index" << std::endl;

		if ( indexfilename.size() )
		{
			libmaus::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Tindex(new libmaus::bambam::BgzfDeflateOutputCallbackBamIndex(tmpfileindex));
			Pindex = UNIQUE_PTR_MOVE(Tindex);
			cbs.push_back(Pindex.get());
		}
	}
	std::vector< ::libmaus::lz::BgzfDeflateOutputCallback * > * Pcbs = 0;
	if ( cbs.size() )
		Pcbs = &cbs;
	/*
	 * end md5/index callbacks
	 */
		
	::libmaus::bambam::BamWriter::unique_ptr_type writer(new ::libmaus::bambam::BamWriter(std::cout,uphead,level,Pcbs));
	libmaus::bambam::BamAlignment & algn = dec.getAlignment();
	uint64_t c = 0;

	while ( dec.readAlignment() )
	{
		bool const ok = split12(algn);
		
		if ( ok )
			algn.serialise(writer->getStream());
 			
		if ( verbose && (++c & (1024*1024-1)) == 0 )
			std::cerr << "[V] " << c/(1024*1024) << std::endl;
	}

	writer.reset();

	if ( Pmd5cb )
	{
		Pmd5cb->saveDigestAsFile(md5filename);
	}
	if ( Pindex )
	{
		Pindex->flush(std::string(indexfilename));
	}
	
	return EXIT_SUCCESS;
}
Esempio n. 10
0
int bammaskflags(::libmaus2::util::ArgInfo const & arginfo)
{
	uint64_t const maskpos = arginfo.getValue<uint64_t>("maskpos",0xFFFFUL);
	uint64_t const maskneg = arginfo.getValue<uint64_t>("maskneg",getDefaultMaskNeg());
	uint64_t const mask = maskpos & (~maskneg);

	if ( mask )
	{
		std::cerr << "Keeping flags ";
		for ( uint64_t i = 1; i <= ::libmaus2::bambam::BamFlagBase::LIBMAUS2_BAMBAM_FSUPPLEMENTARY; i <<= 1 )
			if ( mask & i )
				std::cerr << static_cast< ::libmaus2::bambam::BamFlagBase::bam_flags >(i) << ";";
		std::cerr << std::endl;
		std::cerr << "Erasing flags ";
		for ( uint64_t i = 1; i <= ::libmaus2::bambam::BamFlagBase::LIBMAUS2_BAMBAM_FSUPPLEMENTARY; i <<= 1 )
			if ( !(mask & i) )
				std::cerr << static_cast< ::libmaus2::bambam::BamFlagBase::bam_flags >(i) << ";";
		std::cerr << std::endl;
	}
	else
	{
		std::cerr << "Erasing all flags." << std::endl;
	}

	int const level = libmaus2::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(arginfo.getValue<int>("level",getDefaultLevel()));
	int const resetmatecoord = arginfo.getValue<int>("resetmatecoord",0);

	::libmaus2::bambam::BamDecoder BD(std::cin);
	::libmaus2::bambam::BamHeader const & bamheader = BD.getHeader();

	std::string const headertext(bamheader.text);

	// add PG line to header
	std::string const upheadtext = ::libmaus2::bambam::ProgramHeaderLineSet::addProgramLine(
		headertext,
		"bammaskflags", // ID
		"bammaskflags", // PN
		arginfo.commandline, // CL
		::libmaus2::bambam::ProgramHeaderLineSet(headertext).getLastIdInChain(), // PP
		std::string(PACKAGE_VERSION) // VN
	);
	// construct new header
	::libmaus2::bambam::BamHeader uphead(upheadtext);

	::libmaus2::bambam::BamAlignment & alignment = BD.getAlignment();

	/*
	 * start index/md5 callbacks
	 */
	std::string const tmpfilenamebase = arginfo.getValue<std::string>("tmpfile",arginfo.getDefaultTmpFileName());
	std::string const tmpfileindex = tmpfilenamebase + "_index";
	::libmaus2::util::TempFileRemovalContainer::addTempFile(tmpfileindex);

	std::string md5filename;
	std::string indexfilename;

	std::vector< ::libmaus2::lz::BgzfDeflateOutputCallback * > cbs;
	::libmaus2::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Pmd5cb;
	if ( arginfo.getValue<unsigned int>("md5",getDefaultMD5()) )
	{
		if ( arginfo.hasArg("md5filename") &&  arginfo.getUnparsedValue("md5filename","") != "" )
			md5filename = arginfo.getUnparsedValue("md5filename","");
		else
			std::cerr << "[V] no filename for md5 given, not creating hash" << std::endl;

		if ( md5filename.size() )
		{
			::libmaus2::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Tmd5cb(new ::libmaus2::lz::BgzfDeflateOutputCallbackMD5);
			Pmd5cb = UNIQUE_PTR_MOVE(Tmd5cb);
			cbs.push_back(Pmd5cb.get());
		}
	}
	libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Pindex;
	if ( arginfo.getValue<unsigned int>("index",getDefaultIndex()) )
	{
		if ( arginfo.hasArg("indexfilename") &&  arginfo.getUnparsedValue("indexfilename","") != "" )
			indexfilename = arginfo.getUnparsedValue("indexfilename","");
		else
			std::cerr << "[V] no filename for index given, not creating index" << std::endl;

		if ( indexfilename.size() )
		{
			libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Tindex(new libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex(tmpfileindex));
			Pindex = UNIQUE_PTR_MOVE(Tindex);
			cbs.push_back(Pindex.get());
		}
	}
	std::vector< ::libmaus2::lz::BgzfDeflateOutputCallback * > * Pcbs = 0;
	if ( cbs.size() )
		Pcbs = &cbs;
	/*
	 * end md5/index callbacks
	 */

	::libmaus2::bambam::BamWriter::unique_ptr_type writer(new ::libmaus2::bambam::BamWriter(std::cout,uphead,level,Pcbs));

	while ( BD.readAlignment() )
	{
		alignment.putFlags(alignment.getFlags() & mask);
		if ( resetmatecoord )
		{
			alignment.putNextRefId(-1);
			alignment.putNextPos(-1);
		}
		alignment.serialise(writer->getStream());
	}

	writer.reset();

	if ( Pmd5cb )
	{
		Pmd5cb->saveDigestAsFile(md5filename);
	}
	if ( Pindex )
	{
		Pindex->flush(std::string(indexfilename));
	}

	return EXIT_SUCCESS;
}
Esempio n. 11
0
int main(int argc, char * argv[])
{
	try
	{
		::libmaus2::util::ArgInfo const arginfo(argc,argv);
		
		for ( uint64_t i = 0; i < arginfo.restargs.size(); ++i )
			if ( 
				arginfo.restargs[i] == "-v"
				||
				arginfo.restargs[i] == "--version"
			)
			{
				std::cerr << ::biobambam2::Licensing::license();
				return EXIT_SUCCESS;
			}
			else if ( 
				arginfo.restargs[i] == "-h"
				||
				arginfo.restargs[i] == "--help"
			)
			{
				std::cerr << ::biobambam2::Licensing::license();
				std::cerr << std::endl;
				std::cerr << "Key=Value pairs:" << std::endl;
				std::cerr << std::endl;
				
				std::vector< std::pair<std::string,std::string> > V;
			
				V.push_back ( std::pair<std::string,std::string> ( "level=<["+::biobambam2::Licensing::formatNumber(getDefaultLevel())+"]>", libmaus2::bambam::BamBlockWriterBaseFactory::getBamOutputLevelHelpText() ) );
				V.push_back ( std::pair<std::string,std::string> ( "verbose=<["+::biobambam2::Licensing::formatNumber(getDefaultVerbose())+"]>", "print progress information" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5=<["+::biobambam2::Licensing::formatNumber(getDefaultMD5())+"]>", "create md5 check sum (default: 0)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5filename=<filename>", "file name for md5 check sum" ) );
				V.push_back ( std::pair<std::string,std::string> ( "resetheadertext=[<>]", "replacement SAM header text file (default: filter header in source BAM file)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "exclude=["+getDefaultExcludeFlags()+"]", "drop alignments having any of the given flags set" ) );
				V.push_back ( std::pair<std::string,std::string> ( std::string("resetaux=<[")+::biobambam2::Licensing::formatNumber(getDefaultResetAux())+"]>", "reset auxiliary fields" ) );
				V.push_back ( std::pair<std::string,std::string> ( "auxfilter=[<>]", "comma separated list of aux tags to keep if resetaux=0 (default: keep all)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "resetsortorder=<["+::biobambam2::Licensing::formatNumber(getDefaultResetSortOrder())+"]>", "set sort order to unknown (default: 1)" ) );

				::biobambam2::Licensing::printMap(std::cerr,V);

				std::cerr << std::endl;

				std::cerr << "Alignment flags: PAIRED,PROPER_PAIR,UNMAP,MUNMAP,REVERSE,MREVERSE,READ1,READ2,SECONDARY,QCFAIL,DUP,SUPPLEMENTARY" << std::endl;
				
				return EXIT_SUCCESS;
			}
			
		return bamreset(arginfo);
	}
	catch(std::exception const & ex)
	{
		std::cerr << ex.what() << std::endl;
		return EXIT_FAILURE;
	}
}
Esempio n. 12
0
int main(int argc, char * argv[])
{
	try
	{
		::libmaus2::util::ArgInfo const arginfo(argc,argv);

		for ( uint64_t i = 0; i < arginfo.restargs.size(); ++i )
			if (
				arginfo.restargs[i] == "-v"
				||
				arginfo.restargs[i] == "--version"
			)
			{
				std::cerr << ::biobambam2::Licensing::license();
				return EXIT_SUCCESS;
			}
			else if (
				arginfo.restargs[i] == "-h"
				||
				arginfo.restargs[i] == "--help"
			)
			{
				std::cerr << ::biobambam2::Licensing::license();
				std::cerr << std::endl;
				std::cerr << "Key=Value pairs:" << std::endl;
				std::cerr << std::endl;

				std::vector< std::pair<std::string,std::string> > V;

				V.push_back ( std::pair<std::string,std::string> ( "verbose=<["+::biobambam2::Licensing::formatNumber(getDefaultVerbose())+"]>", "print stats at the end of a successfull run" ) );
				V.push_back ( std::pair<std::string,std::string> ( "basequalhist=<["+::biobambam2::Licensing::formatNumber(getDefaultBaseQualHist())+"]>", "print base quality histogram at end of a successfull run" ) );
				V.push_back ( std::pair<std::string,std::string> ( "passthrough=<["+::biobambam2::Licensing::formatNumber(getDefaultPassThrough())+"]>", "write alignments to standard output (default: do not pass through)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "level=<["+::biobambam2::Licensing::formatNumber(getDefaultLevel())+"]>", libmaus2::bambam::BamBlockWriterBaseFactory::getBamOutputLevelHelpText() ) );
				V.push_back ( std::pair<std::string,std::string> ( "tmpfile=<filename>", "prefix for temporary files, default: create files in current directory (passthrough=1, index=1 only)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5=<["+::biobambam2::Licensing::formatNumber(getDefaultMD5())+"]>", "create md5 check sum (default: 0, passthrough=1 only)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5filename=<filename>", "file name for md5 check sum" ) );
				V.push_back ( std::pair<std::string,std::string> ( "index=<["+::biobambam2::Licensing::formatNumber(getDefaultIndex())+"]>", "create BAM index (default: 0, passthrough=1 only)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "indexfilename=<filename>", "file name for BAM index file" ) );
				V.push_back ( std::pair<std::string,std::string> ( std::string("inputformat=<[")+getDefaultInputFormat()+"]>", std::string("input format (") + libmaus2::bambam::BamMultiAlignmentDecoderFactory::getValidInputFormats() + ")" ) );
				V.push_back ( std::pair<std::string,std::string> ( std::string("outputformat=<[")+libmaus2::bambam::BamBlockWriterBaseFactory::getDefaultOutputFormat()+"]>", std::string("output format (") + libmaus2::bambam::BamBlockWriterBaseFactory::getValidOutputFormats() + ", passthrough=1 only)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "I=<[stdin]>", "input filename (standard input if unset)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "inputthreads=<[1]>", "input helper threads (for inputformat=bam only, default: 1)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "reference=<>", "reference FastA (.fai file required, for cram i/o only)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "range=<>", "coordinate range to be processed (for coordinate sorted indexed BAM input only)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "outputthreads=<[1]>", "output helper threads (for outputformat=bam only, default: 1, passthrough=1 only)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "O=<[stdout]>", "output filename (standard output if unset, passthrough=1 only)" ) );

				::biobambam2::Licensing::printMap(std::cerr,V);

				std::cerr << std::endl;
				return EXIT_SUCCESS;
			}


		return bamvalidate(arginfo);
	}
	catch(std::exception const & ex)
	{
		std::cerr << ex.what() << std::endl;
		return EXIT_FAILURE;
	}

}
Esempio n. 13
0
int main(int argc, char * argv[])
{
	try
	{
		libmaus2::util::ArgInfo const arginfo(argc,argv);

		for ( uint64_t i = 0; i < arginfo.restargs.size(); ++i )
			if (
				arginfo.restargs[i] == "-v"
				||
				arginfo.restargs[i] == "--version"
			)
			{
				std::cerr << ::biobambam2::Licensing::license();
				return EXIT_SUCCESS;
			}
			else if (
				arginfo.restargs[i] == "-h"
				||
				arginfo.restargs[i] == "--help"
			)
			{
				std::cerr << ::biobambam2::Licensing::license();
				std::cerr << std::endl;
				std::cerr << "Key=Value pairs:" << std::endl;
				std::cerr << std::endl;

				std::vector< std::pair<std::string,std::string> > V;

				V.push_back ( std::pair<std::string,std::string> ( "level=<["+::biobambam2::Licensing::formatNumber(getDefaultLevel())+"]>", libmaus2::bambam::BamBlockWriterBaseFactory::getBamOutputLevelHelpText() ) );
				V.push_back ( std::pair<std::string,std::string> ( "verbose=<["+::biobambam2::Licensing::formatNumber(getDefaultVerbose())+"]>", "print progress report" ) );
				V.push_back ( std::pair<std::string,std::string> ( std::string("inputformat=<[")+getDefaultInputFormat()+"]>", std::string("input format (") + libmaus2::bambam::BamMultiAlignmentDecoderFactory::getValidInputFormats() + ")" ) );
				V.push_back ( std::pair<std::string,std::string> ( std::string("outputformat=<[")+libmaus2::bambam::BamBlockWriterBaseFactory::getDefaultOutputFormat()+"]>", std::string("output format (") + libmaus2::bambam::BamBlockWriterBaseFactory::getValidOutputFormats() + ")" ) );
				V.push_back ( std::pair<std::string,std::string> ( "I=<[stdin]>", "input filename (standard input if unset)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "inputthreads=<[1]>", "input helper threads (for inputformat=bam only, default: 1)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "reference=<>", "reference FastA (.fai file required, for cram i/o only)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "range=<>", "coordinate range to be processed (for coordinate sorted indexed BAM input only)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "outputthreads=<[1]>", "output helper threads (for outputformat=bam only, default: 1)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "O=<[stdout]>", "output filename (standard output if unset)" ) );
				V.push_back ( std::pair<std::string,std::string> ( std::string("prefix=<[")+getDefaultPrefix()+"]>", "prefix of output file names" ) );
				V.push_back ( std::pair<std::string,std::string> ( "thres=<["+::biobambam2::Licensing::formatNumber(getDefaultSizeThres())+"]>", "size threshold for the creation of next file" ) );

				::biobambam2::Licensing::printMap(std::cerr,V);

				std::cerr << std::endl;
				return EXIT_SUCCESS;
			}

		return bamexplode(arginfo);
	}
	catch(std::exception const & ex)
	{
		std::cerr << ex.what() << std::endl;
		return EXIT_FAILURE;
	}
}
Esempio n. 14
0
int bamsort(::libmaus::util::ArgInfo const & arginfo)
{
	::libmaus::util::TempFileRemovalContainer::setup();
	
	bool const inputisstdin = (!arginfo.hasArg("I")) || (arginfo.getUnparsedValue("I","-") == "-");
	bool const outputisstdout = (!arginfo.hasArg("O")) || (arginfo.getUnparsedValue("O","-") == "-");

	if ( isatty(STDIN_FILENO) && inputisstdin && (arginfo.getValue<std::string>("inputformat","bam") != "sam") )
	{
		::libmaus::exception::LibMausException se;
		se.getStream() << "Refusing to read binary data from terminal, please redirect standard input to pipe or file." << std::endl;
		se.finish();
		throw se;
	}

	if ( isatty(STDOUT_FILENO) && outputisstdout && (arginfo.getValue<std::string>("outputformat","bam") != "sam") )
	{
		::libmaus::exception::LibMausException se;
		se.getStream() << "Refusing write binary data to terminal, please redirect standard output to pipe or file." << std::endl;
		se.finish();
		throw se;
	}

	int const verbose = arginfo.getValue<int>("verbose",getDefaultVerbose());
	bool const disablevalidation = arginfo.getValue<int>("disablevalidation",getDefaultDisableValidation());

	std::string const inputformat = arginfo.getUnparsedValue("inputformat",getDefaultInputFormat());
	int const level = arginfo.getValue<int>("level",getDefaultLevel());
	switch ( level )
	{
		case Z_NO_COMPRESSION:
		case Z_BEST_SPEED:
		case Z_BEST_COMPRESSION:
		case Z_DEFAULT_COMPRESSION:
			break;
		default:
		{
			::libmaus::exception::LibMausException se;
			se.getStream()
				<< "Unknown compression level, please use"
				<< " level=" << Z_DEFAULT_COMPRESSION << " (default) or"
				<< " level=" << Z_BEST_SPEED << " (fast) or"
				<< " level=" << Z_BEST_COMPRESSION << " (best) or"
				<< " level=" << Z_NO_COMPRESSION << " (no compression)" << std::endl;
			se.finish();
			throw se;
		}
			break;
	}

	// prefix for tmp files
	std::string const tmpfilenamebase = arginfo.getValue<std::string>("tmpfile",arginfo.getDefaultTmpFileName());
	std::string const tmpfilenameout = tmpfilenamebase + "_bamsort";
	::libmaus::util::TempFileRemovalContainer::addTempFile(tmpfilenameout);
	uint64_t blockmem = arginfo.getValue<uint64_t>("blockmb",getDefaultBlockSize())*1024*1024;
	std::string const sortorder = arginfo.getValue<std::string>("SO","coordinate");
	bool const fixmates = arginfo.getValue<int>("fixmates",getDefaultFixMates());
	uint64_t sortthreads = arginfo.getValue<uint64_t>("sortthreads",getDefaultSortThreads());

	// input decoder wrapper
	libmaus::bambam::BamAlignmentDecoderWrapper::unique_ptr_type decwrapper(
		libmaus::bambam::BamMultiAlignmentDecoderFactory::construct(
			arginfo,false // put rank
		)
	);
	::libmaus::bambam::BamAlignmentDecoder * ppdec = &(decwrapper->getDecoder());
	::libmaus::bambam::BamAlignmentDecoder & dec = *ppdec;
	if ( disablevalidation )
		dec.disableValidation();
	::libmaus::bambam::BamHeader const & header = dec.getHeader();

	std::string const headertext(header.text);

	// add PG line to header
	std::string const upheadtext = ::libmaus::bambam::ProgramHeaderLineSet::addProgramLine(
		headertext,
		"bamsort", // ID
		"bamsort", // PN
		arginfo.commandline, // CL
		::libmaus::bambam::ProgramHeaderLineSet(headertext).getLastIdInChain(), // PP
		std::string(PACKAGE_VERSION) // VN			
	);
	// construct new header
	::libmaus::bambam::BamHeader uphead(upheadtext);

	/*
	 * start index/md5 callbacks
	 */
	std::string const tmpfileindex = tmpfilenamebase + "_index";
	::libmaus::util::TempFileRemovalContainer::addTempFile(tmpfileindex);

	std::string md5filename;
	std::string indexfilename;

	std::vector< ::libmaus::lz::BgzfDeflateOutputCallback * > cbs;
	::libmaus::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Pmd5cb;
	if ( arginfo.getValue<unsigned int>("md5",getDefaultMD5()) )
	{
		if ( arginfo.hasArg("md5filename") &&  arginfo.getUnparsedValue("md5filename","") != "" )
			md5filename = arginfo.getUnparsedValue("md5filename","");
		else
			std::cerr << "[V] no filename for md5 given, not creating hash" << std::endl;

		if ( md5filename.size() )
		{
			::libmaus::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Tmd5cb(new ::libmaus::lz::BgzfDeflateOutputCallbackMD5);
			Pmd5cb = UNIQUE_PTR_MOVE(Tmd5cb);
			cbs.push_back(Pmd5cb.get());
		}
	}
	libmaus::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Pindex;
	if ( arginfo.getValue<unsigned int>("index",getDefaultIndex()) )
	{
		if ( arginfo.hasArg("indexfilename") &&  arginfo.getUnparsedValue("indexfilename","") != "" )
			indexfilename = arginfo.getUnparsedValue("indexfilename","");
		else
			std::cerr << "[V] no filename for index given, not creating index" << std::endl;

		if ( indexfilename.size() )
		{
			libmaus::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Tindex(new libmaus::bambam::BgzfDeflateOutputCallbackBamIndex(tmpfileindex));
			Pindex = UNIQUE_PTR_MOVE(Tindex);
			cbs.push_back(Pindex.get());
		}
	}
	std::vector< ::libmaus::lz::BgzfDeflateOutputCallback * > * Pcbs = 0;
	if ( cbs.size() )
		Pcbs = &cbs;
	/*
	 * end md5/index callbacks
	 */
	if ( sortorder != "queryname" )
		uphead.changeSortOrder("coordinate");
	else
		uphead.changeSortOrder("queryname");

	libmaus::bambam::BamBlockWriterBase::unique_ptr_type Pout ( libmaus::bambam::BamBlockWriterBaseFactory::construct(uphead, arginfo, Pcbs) );

	if ( fixmates )
	{
		if ( sortorder != "queryname" )
		{
			::libmaus::bambam::BamEntryContainer< ::libmaus::bambam::BamAlignmentPosComparator > 
				BEC(blockmem,tmpfilenameout,sortthreads);

			if ( verbose )
				std::cerr << "[V] Reading alignments from source." << std::endl;
			uint64_t incnt = 0;

			// current alignment
			libmaus::bambam::BamAlignment & curalgn = dec.getAlignment();
			// previous alignment
			libmaus::bambam::BamAlignment prevalgn;
			// previous alignment valid
			bool prevalgnvalid = false;
			// MQ field filter
			libmaus::bambam::BamAuxFilterVector MQfilter;
			MQfilter.set("MQ");
			
			while ( dec.readAlignment() )
			{
				if ( curalgn.isSecondary() || curalgn.isSupplementary() )
				{
					BEC.putAlignment(curalgn);
				}
				else if ( prevalgnvalid )
				{
					// different name
					if ( strcmp(curalgn.getName(),prevalgn.getName()) )
					{
						BEC.putAlignment(prevalgn);
						curalgn.swap(prevalgn);
					}
					// same name
					else
					{
						libmaus::bambam::BamAlignment::fixMateInformation(prevalgn,curalgn,MQfilter);
						BEC.putAlignment(prevalgn);
						BEC.putAlignment(curalgn);
						prevalgnvalid = false;
					}
				}
				else
				{
					prevalgn.swap(curalgn);
					prevalgnvalid = true;
				}
				
				if ( verbose && ( ( ++incnt & ((1ull<<20)-1) ) == 0 ) )
					std::cerr << "[V] " << incnt << std::endl;
			}
			
			if ( prevalgnvalid )
			{
				BEC.putAlignment(prevalgn);
				prevalgnvalid = false;
			}

			if ( verbose )
				std::cerr << "[V] read " << incnt << " alignments" << std::endl;

			// BEC.createOutput(std::cout, uphead, level, verbose, Pcbs);
			BEC.createOutput(*Pout, verbose);
		}
		else
		{
			::libmaus::bambam::BamEntryContainer< ::libmaus::bambam::BamAlignmentNameComparator > 
				BEC(blockmem,tmpfilenameout,sortthreads);
			
			if ( verbose )
				std::cerr << "[V] Reading alignments from source." << std::endl;
			uint64_t incnt = 0;
			
			// current alignment
			libmaus::bambam::BamAlignment & curalgn = dec.getAlignment();
			// previous alignment
			libmaus::bambam::BamAlignment prevalgn;
			// previous alignment valid
			bool prevalgnvalid = false;
			// MQ field filter
			libmaus::bambam::BamAuxFilterVector MQfilter;
			MQfilter.set("MQ");
			
			while ( dec.readAlignment() )
			{
				if ( curalgn.isSecondary() || curalgn.isSupplementary() )
				{
					BEC.putAlignment(curalgn);
				}
				else if ( prevalgnvalid )
				{
					// different name
					if ( strcmp(curalgn.getName(),prevalgn.getName()) )
					{
						BEC.putAlignment(prevalgn);
						curalgn.swap(prevalgn);
					}
					// same name
					else
					{
						libmaus::bambam::BamAlignment::fixMateInformation(prevalgn,curalgn,MQfilter);
						BEC.putAlignment(prevalgn);
						BEC.putAlignment(curalgn);
						prevalgnvalid = false;
					}
				}
				else
				{
					prevalgn.swap(curalgn);
					prevalgnvalid = true;
				}
				
				if ( verbose && ( ( ++incnt & ((1ull<<20)-1) ) == 0 ) )
					std::cerr << "[V] " << incnt << std::endl;
			}
			
			if ( prevalgnvalid )
			{
				BEC.putAlignment(prevalgn);
				prevalgnvalid = false;
			}
			
			if ( verbose )
				std::cerr << "[V] read " << incnt << " alignments" << std::endl;

			// BEC.createOutput(std::cout, uphead, level, verbose, Pcbs);
			BEC.createOutput(*Pout, verbose);
		}
	}
	else
	{
		if ( sortorder != "queryname" )
		{
			::libmaus::bambam::BamEntryContainer< ::libmaus::bambam::BamAlignmentPosComparator > BEC(blockmem,tmpfilenameout,sortthreads);

			if ( verbose )
				std::cerr << "[V] Reading alignments from source." << std::endl;
			uint64_t incnt = 0;
			
			while ( dec.readAlignment() )
			{
				BEC.putAlignment(dec.getAlignment());
				incnt++;
				if ( verbose && (incnt % (1024*1024) == 0) )
					std::cerr << "[V] " << incnt/(1024*1024) << "M" << std::endl;
			}

			if ( verbose )
				std::cerr << "[V] read " << incnt << " alignments" << std::endl;

			// BEC.createOutput(std::cout, uphead, level, verbose, Pcbs);
			BEC.createOutput(*Pout, verbose);
		}
		else
		{
			::libmaus::bambam::BamEntryContainer< ::libmaus::bambam::BamAlignmentNameComparator > BEC(blockmem,tmpfilenameout,sortthreads);
			
			if ( verbose )
				std::cerr << "[V] Reading alignments from source." << std::endl;
			uint64_t incnt = 0;
			
			while ( dec.readAlignment() )
			{
				BEC.putAlignment(dec.getAlignment());
				incnt++;
				if ( verbose && (incnt % (1024*1024) == 0) )
					std::cerr << "[V] " << incnt/(1024*1024) << "M" << std::endl;
			}
			
			if ( verbose )
				std::cerr << "[V] read " << incnt << " alignments" << std::endl;

			// BEC.createOutput(std::cout, uphead, level, verbose, Pcbs);
			BEC.createOutput(*Pout, verbose);
		}
	}

	// flush encoder so callbacks see all output data
	Pout.reset();

	if ( Pmd5cb )
	{
		Pmd5cb->saveDigestAsFile(md5filename);
	}
	if ( Pindex )
	{
		Pindex->flush(std::string(indexfilename));
	}

	return EXIT_SUCCESS;
}
Esempio n. 15
0
void elog::setLevel(enum level _level) {
	getDefaultLevel() = _level;
	for (size_t iii = 0; iii < getList().size(); ++iii) {
		getList()[iii].second = _level;
	}
}
Esempio n. 16
0
int bamcat(libmaus::util::ArgInfo const & arginfo)
{
	if ( isatty(STDOUT_FILENO) )
	{
		::libmaus::exception::LibMausException se;
		se.getStream() << "Refusing write binary data to terminal, please redirect standard output to pipe or file." << std::endl;
		se.finish();
		throw se;
	}

	int const level = arginfo.getValue<int>("level",getDefaultLevel());
	int const verbose = arginfo.getValue<int>("verbose",getDefaultVerbose());
	
	switch ( level )
	{
		case Z_NO_COMPRESSION:
		case Z_BEST_SPEED:
		case Z_BEST_COMPRESSION:
		case Z_DEFAULT_COMPRESSION:
			break;
		default:
		{
			::libmaus::exception::LibMausException se;
			se.getStream()
				<< "Unknown compression level, please use"
				<< " level=" << Z_DEFAULT_COMPRESSION << " (default) or"
				<< " level=" << Z_BEST_SPEED << " (fast) or"
				<< " level=" << Z_BEST_COMPRESSION << " (best) or"
				<< " level=" << Z_NO_COMPRESSION << " (no compression)" << std::endl;
			se.finish();
			throw se;
		}
			break;
	}

	std::vector<std::string> inputfilenames = arginfo.getPairValues("I");
	
	for ( uint64_t i = 0; i < arginfo.restargs.size(); ++i )
		inputfilenames.push_back(arginfo.restargs[i]);
	
	libmaus::bambam::BamCat bamdec(inputfilenames /* ,true */);
	libmaus::bambam::BamAlignment const & algn = bamdec.getAlignment();
	libmaus::bambam::BamHeader const & header = bamdec.getHeader();
	::libmaus::bambam::BamHeader::unique_ptr_type uphead(updateHeader(arginfo,header));

	/*
	 * start index/md5 callbacks
	 */
	std::string const tmpfilenamebase = arginfo.getValue<std::string>("tmpfile",arginfo.getDefaultTmpFileName());
	std::string const tmpfileindex = tmpfilenamebase + "_index";
	::libmaus::util::TempFileRemovalContainer::addTempFile(tmpfileindex);

	std::string md5filename;
	std::string indexfilename;

	std::vector< ::libmaus::lz::BgzfDeflateOutputCallback * > cbs;
	::libmaus::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Pmd5cb;
	if ( arginfo.getValue<unsigned int>("md5",getDefaultMD5()) )
	{
		if ( arginfo.hasArg("md5filename") &&  arginfo.getUnparsedValue("md5filename","") != "" )
			md5filename = arginfo.getUnparsedValue("md5filename","");
		else
			std::cerr << "[V] no filename for md5 given, not creating hash" << std::endl;

		if ( md5filename.size() )
		{
			::libmaus::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Tmd5cb(new ::libmaus::lz::BgzfDeflateOutputCallbackMD5);
			Pmd5cb = UNIQUE_PTR_MOVE(Tmd5cb);
			cbs.push_back(Pmd5cb.get());
		}
	}
	libmaus::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Pindex;
	if ( arginfo.getValue<unsigned int>("index",getDefaultIndex()) )
	{
		if ( arginfo.hasArg("indexfilename") &&  arginfo.getUnparsedValue("indexfilename","") != "" )
			indexfilename = arginfo.getUnparsedValue("indexfilename","");
		else
			std::cerr << "[V] no filename for index given, not creating index" << std::endl;

		if ( indexfilename.size() )
		{
			libmaus::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Tindex(new libmaus::bambam::BgzfDeflateOutputCallbackBamIndex(tmpfileindex));
			Pindex = UNIQUE_PTR_MOVE(Tindex);
			cbs.push_back(Pindex.get());
		}
	}
	std::vector< ::libmaus::lz::BgzfDeflateOutputCallback * > * Pcbs = 0;
	if ( cbs.size() )
		Pcbs = &cbs;
	/*
	 * end md5/index callbacks
	 */


	::libmaus::bambam::BamWriter::unique_ptr_type writer(new ::libmaus::bambam::BamWriter(std::cout,*uphead,level,Pcbs));
	libmaus::bambam::BamWriter::stream_type & bamoutstr = writer->getStream();
	if ( verbose )
	{
		uint64_t c = 0;
		while ( bamdec.readAlignment() )
		{
			algn.serialise(bamoutstr);
			
			if ( ((++c) & ((1ull<<20)-1)) == 0 )
				std::cerr << "[V] " << c << std::endl;
		}
		
		std::cerr << "[V] " << c << std::endl;
	}
	else
		while ( bamdec.readAlignment() )
			algn.serialise(bamoutstr);

	writer.reset();

	if ( Pmd5cb )
	{
		Pmd5cb->saveDigestAsFile(md5filename);
	}
	if ( Pindex )
	{
		Pindex->flush(std::string(indexfilename));
	}

	return EXIT_SUCCESS;
}
int bamfixmatecoordinatesnamesorted(::libmaus::util::ArgInfo const & arginfo)
{
	bool const verbose = arginfo.getValue<unsigned int>("verbose",getDefaultVerbose());
	
	::libmaus::timing::RealTimeClock rtc; rtc.start();
	
	// gzip compression level for output
	int const level = libmaus::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(arginfo.getValue<int>("level",getDefaultLevel()));
	
	::libmaus::bambam::BamDecoder bamfile(std::cin);
	std::string const headertext(bamfile.getHeader().text);

	// add PG line to header
	std::string const upheadtext = ::libmaus::bambam::ProgramHeaderLineSet::addProgramLine(
		headertext,
		"bamfixmatecoordinatesnamesorted", // ID
		"bamfixmatecoordinatesnamesorted", // PN
		arginfo.commandline, // CL
		::libmaus::bambam::ProgramHeaderLineSet(headertext).getLastIdInChain(), // PP
		std::string(PACKAGE_VERSION) // VN			
	);
	// construct new header
	::libmaus::bambam::BamHeader uphead(upheadtext);
	
	if ( uphead.getSortOrder() != "queryname" )
		uphead.changeSortOrder("unknown");
		
	std::string const & finalheadtext = uphead.text;
	::libmaus::bambam::BamHeader finalheader(finalheadtext);

	/*
	 * start index/md5 callbacks
	 */
	std::string const tmpfilenamebase = arginfo.getValue<std::string>("tmpfile",arginfo.getDefaultTmpFileName());
	std::string const tmpfileindex = tmpfilenamebase + "_index";
	::libmaus::util::TempFileRemovalContainer::addTempFile(tmpfileindex);

	std::string md5filename;
	std::string indexfilename;

	std::vector< ::libmaus::lz::BgzfDeflateOutputCallback * > cbs;
	::libmaus::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Pmd5cb;
	if ( arginfo.getValue<unsigned int>("md5",getDefaultMD5()) )
	{
		if ( arginfo.hasArg("md5filename") &&  arginfo.getUnparsedValue("md5filename","") != "" )
			md5filename = arginfo.getUnparsedValue("md5filename","");
		else
			std::cerr << "[V] no filename for md5 given, not creating hash" << std::endl;

		if ( md5filename.size() )
		{
			::libmaus::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Tmd5cb(new ::libmaus::lz::BgzfDeflateOutputCallbackMD5);
			Pmd5cb = UNIQUE_PTR_MOVE(Tmd5cb);
			cbs.push_back(Pmd5cb.get());
		}
	}
	libmaus::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Pindex;
	if ( arginfo.getValue<unsigned int>("index",getDefaultIndex()) )
	{
		if ( arginfo.hasArg("indexfilename") &&  arginfo.getUnparsedValue("indexfilename","") != "" )
			indexfilename = arginfo.getUnparsedValue("indexfilename","");
		else
			std::cerr << "[V] no filename for index given, not creating index" << std::endl;

		if ( indexfilename.size() )
		{
			libmaus::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Tindex(new libmaus::bambam::BgzfDeflateOutputCallbackBamIndex(tmpfileindex));
			Pindex = UNIQUE_PTR_MOVE(Tindex);
			cbs.push_back(Pindex.get());
		}
	}
	std::vector< ::libmaus::lz::BgzfDeflateOutputCallback * > * Pcbs = 0;
	if ( cbs.size() )
		Pcbs = &cbs;
	/*
	 * end md5/index callbacks
	 */

	::libmaus::bambam::BamWriter::unique_ptr_type writer(new ::libmaus::bambam::BamWriter(std::cout,finalheader,level,Pcbs));
	std::pair< std::pair< ::libmaus::bambam::BamAlignment::shared_ptr_type, bool> , std::pair< ::libmaus::bambam::BamAlignment::shared_ptr_type, bool> > 
		P(std::pair< ::libmaus::bambam::BamAlignment::shared_ptr_type, bool>(::libmaus::bambam::BamAlignment::shared_ptr_type(),false),std::pair< ::libmaus::bambam::BamAlignment::shared_ptr_type, bool>(::libmaus::bambam::BamAlignment::shared_ptr_type(),false));
	
	// try to read two alignments	
	P.first.second  = bamfile.readAlignment();
	if ( P.first.second )
	{
		P.first.first   = bamfile.salignment();
		P.second.second = P.first.second && bamfile.readAlignment();
		P.second.first  = bamfile.salignment();
	}
	
	uint64_t single = 0, pairs = 0;
	uint64_t proc = 0;
	uint64_t lastproc = 0;
	uint64_t const mod = 1024*1024;
	
	// while we have two alignments
	while ( P.first.second && P.second.second )
	{
		uint32_t const aflags = P.first.first->getFlags();
		uint32_t const bflags = P.second.first->getFlags();
	
		// same name?
		if ( 
			(aflags & ::libmaus::bambam::BamFlagBase::LIBMAUS_BAMBAM_FPAIRED)
			&&
			(bflags & ::libmaus::bambam::BamFlagBase::LIBMAUS_BAMBAM_FPAIRED)
			&&
			(! strcmp(P.first.first->getName(),P.second.first->getName()))
		)
		{			
			unsigned int const amap = (aflags & ::libmaus::bambam::BamFlagBase::LIBMAUS_BAMBAM_FUNMAP) ? 0 : 1;
			unsigned int const bmap = (bflags & ::libmaus::bambam::BamFlagBase::LIBMAUS_BAMBAM_FUNMAP) ? 0 : 1;

			// std::cerr << "Pair " << bam1_qname(P.first.first->alignment) << " amap=" << amap << " bmap=" << bmap << std::endl;
			
			// if exactly one of the two is mapped
			if ( amap + bmap == 1 )
			{
				::libmaus::bambam::BamAlignment::shared_ptr_type mapped = amap ? P.first.first : P.second.first;
				int64_t const tid = mapped->getRefID();
				int64_t const pos = mapped->getPos();
				
				// std::cerr << "tid=" << tid << " pos=" << pos << std::endl;
				
				// set all tid and pos values
				P.first.first->putRefId(tid);
				P.first.first->putPos(pos);
				P.first.first->putNextRefId(tid);
				P.first.first->putNextPos(pos);
				P.second.first->putRefId(tid);
				P.second.first->putPos(pos);
				P.second.first->putNextRefId(tid);
				P.second.first->putNextPos(pos);
			}
		
			// write alignments
			P.first.first->serialise(writer->getStream());
			P.second.first->serialise(writer->getStream());
			// read new alignments
			P.first.second = bamfile.readAlignment();
			if ( P.first.second )
			{
				P.first.first = bamfile.salignment();
				P.second.second = bamfile.readAlignment();
				P.second.first = bamfile.salignment();
			}
			
			pairs++;
			proc += 2;
		}
		// different names
		else
		{
			// write first alignment
			P.first.first->serialise(writer->getStream());
			// move second to first
			std::swap(P.first,P.second);
			// read new second
			P.second.second = P.first.second && bamfile.readAlignment();
			if ( P.second.second )
				P.second.first = bamfile.salignment();
			
			single++;
			proc += 1;
		}
		
		if ( verbose && (proc/mod != lastproc/mod) )
		{
			std::cerr << proc << "\t" << single << "\t" << pairs << "\t" <<
				proc/rtc.getElapsedSeconds() << "al/s"
				<< std::endl;
			lastproc = proc;
		}
	}
	
	if ( P.first.second )
	{
		P.first.first->serialise(writer->getStream());
		single++;
		proc += 1;
	}

	if ( verbose )
		std::cerr << proc << "\t" << single << "\t" << pairs << "\t" <<
			proc/rtc.getElapsedSeconds() << "al/s"
			<< std::endl;
		
	assert ( ! P.second.second );

	writer.reset();

	if ( Pmd5cb )
	{
		Pmd5cb->saveDigestAsFile(md5filename);
	}	
	if ( Pindex )
	{
		Pindex->flush(std::string(indexfilename));
	}

	return EXIT_SUCCESS;
}
Esempio n. 18
0
int main(int argc, char * argv[])
{
	try
	{
		::libmaus2::util::ArgInfo const arginfo(argc,argv);

		for ( uint64_t i = 0; i < arginfo.restargs.size(); ++i )
			if (
				arginfo.restargs[i] == "-v"
				||
				arginfo.restargs[i] == "--version"
			)
			{
				std::cerr << ::biobambam2::Licensing::license();
				return EXIT_SUCCESS;
			}
			else if (
				arginfo.restargs[i] == "-h"
				||
				arginfo.restargs[i] == "--help"
			)
			{
				std::cerr << ::biobambam2::Licensing::license();
				std::cerr << std::endl;
				std::cerr << "Key=Value pairs:" << std::endl;
				std::cerr << std::endl;

				std::vector< std::pair<std::string,std::string> > V;

				V.push_back ( std::pair<std::string,std::string> ( "level=<["+::biobambam2::Licensing::formatNumber(getDefaultLevel())+"]>", libmaus2::bambam::BamBlockWriterBaseFactory::getBamOutputLevelHelpText() ) );
				V.push_back ( std::pair<std::string,std::string> ( "verbose=<["+::biobambam2::Licensing::formatNumber(getDefaultVerbose())+"]>", "print progress report" ) );
				V.push_back ( std::pair<std::string,std::string> ( "mod=<["+::biobambam2::Licensing::formatNumber(getDefaultMod())+"]>", "print progress for every mod'th alignment if verbose" ) );
				V.push_back ( std::pair<std::string,std::string> ( "ranksplit=<["+::biobambam2::Licensing::formatNumber(getDefaultRankSplit())+"]>", "split rank pairs in names (see bam12split command)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "rankstrip=<["+::biobambam2::Licensing::formatNumber(getDefaultRankStrip())+"]>", "strip ranks of names (see bam12strip command)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "clipreinsert=<["+::biobambam2::Licensing::formatNumber(getDefaultClipReinsert())+"]>", "reinsert clipped sequence fragments (see bamclipreinsert command)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "zztoname=<["+::biobambam2::Licensing::formatNumber(getDefaultZZToName())+"]>", "move rank from zz to name aux field to name (see bamzztoname command)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5=<["+::biobambam2::Licensing::formatNumber(getDefaultMD5())+"]>", "create md5 check sum (default: 0)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "md5filename=<filename>", "file name for md5 check sum (default: extend output file name)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "index=<["+::biobambam2::Licensing::formatNumber(getDefaultIndex())+"]>", "create BAM index (default: 0)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "indexfilename=<filename>", "file name for BAM index file (default: extend output file name)" ) );
				V.push_back ( std::pair<std::string,std::string> ( "tmpfile=<filename>", "prefix for temporary files, default: create files in current directory" ) );
    	    	    	    	V.push_back ( std::pair<std::string,std::string> ( "sanity=<["+::biobambam2::Licensing::formatNumber(getDefaultSanity())+"]>", "extra checking of reads" ) );

				::biobambam2::Licensing::printMap(std::cerr,V);

				std::cerr << std::endl;
				return EXIT_SUCCESS;
			}

		return bam12auxmerge(arginfo);
	}
	catch(std::exception const & ex)
	{
		std::cerr << ex.what() << std::endl;
		return EXIT_FAILURE;
	}
}
Esempio n. 19
0
static void filterBamUsedSequences(
	libmaus::util::ArgInfo const & arginfo,
	std::istream & in,
	::libmaus::bitio::IndexedBitVector const & IBV,
	std::ostream & out
)
{
	libmaus::lz::BgzfInflateStream bgzfin(in);
	libmaus::bambam::BamHeaderLowMem::unique_ptr_type PBHLM ( libmaus::bambam::BamHeaderLowMem::constructFromBAM(bgzfin));

	bool const verbose = arginfo.getValue<unsigned int>("verbose",getDefaultVerbose());
	std::vector< ::libmaus::lz::BgzfDeflateOutputCallback * > cbs;
	::libmaus::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Pmd5cb;
	std::string md5filename;
	if ( arginfo.getValue<unsigned int>("md5",getDefaultMD5()) )
	{
		if ( arginfo.hasArg("md5filename") &&  arginfo.getUnparsedValue("md5filename","") != "" )
			md5filename = arginfo.getUnparsedValue("md5filename","");
		else
			std::cerr << "[V] no filename for md5 given, not creating hash" << std::endl;

		if ( md5filename.size() )
		{
			::libmaus::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Tmd5cb(new ::libmaus::lz::BgzfDeflateOutputCallbackMD5);
			Pmd5cb = UNIQUE_PTR_MOVE(Tmd5cb);
			cbs.push_back(Pmd5cb.get());
		}
	}

	int const level = libmaus::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(arginfo.getValue<int>("level",getDefaultLevel()));
	libmaus::lz::BgzfDeflate<std::ostream>::unique_ptr_type Pbgzfout(
		new libmaus::lz::BgzfDeflate<std::ostream>(
			out,level
		)
	);
	libmaus::lz::BgzfDeflate<std::ostream> & bgzfout = *Pbgzfout;
	
	if ( verbose )
		std::cerr << "[V] writing filtered header...";
	PBHLM->serialiseSequenceSubset(bgzfout,IBV,"bamfilterheader2" /* id */,"bamfilterheader2" /* pn */,
		arginfo.commandline /* pgCL */, PACKAGE_VERSION /* pgVN */
	);
	if ( verbose )
		std::cerr << "done." << std::endl;

	::libmaus::bambam::BamAlignment algn;
	uint64_t c = 0;
	while ( libmaus::bambam::BamAlignmentDecoder::readAlignmentGz(bgzfin,algn) )
	{
		if ( algn.isMapped() )
		{
			int64_t const refid = algn.getRefID();
			assert ( refid >= 0 );
			assert ( IBV.get(refid) );
			algn.putRefId(IBV.rank1(refid)-1);
		}
		else
		{
			algn.putRefId(-1);
		}
		
		if ( algn.isPaired() && algn.isMapped() )
		{
			int64_t const refid = algn.getNextRefID();
			assert ( refid >= 0 );
			assert ( IBV.get(refid) );
			algn.putNextRefId(IBV.rank1(refid)-1);
		}
		else
		{
			algn.putNextRefId(-1);
		}
		
		algn.serialise(bgzfout);
		
		if ( verbose && ( ((++c) & (1024*1024-1)) == 0 ) )
			std::cerr << "[V] " << c/(1024*1024) << std::endl;
	}
	
	bgzfout.flush();
	bgzfout.addEOFBlock();	
		
	Pbgzfout.reset();

	if ( Pmd5cb )
		Pmd5cb->saveDigestAsFile(md5filename);
}
Esempio n. 20
0
int bam12auxmerge(::libmaus2::util::ArgInfo const & arginfo)
{
	if ( isatty(STDIN_FILENO) )
	{
		::libmaus2::exception::LibMausException se;
		se.getStream() << "Refusing to read binary data from terminal, please redirect standard input to pipe or file." << std::endl;
		se.finish();
		throw se;
	}

	if ( isatty(STDOUT_FILENO) )
	{
		::libmaus2::exception::LibMausException se;
		se.getStream() << "Refusing write binary data to terminal, please redirect standard output to pipe or file." << std::endl;
		se.finish();
		throw se;
	}

	std::string const prefilename = arginfo.getRestArg<std::string>(0);
	libmaus2::bambam::BamDecoder bampredec(prefilename);

	int const level = libmaus2::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(arginfo.getValue<int>("level",getDefaultLevel()));
	int const verbose = arginfo.getValue<int>("verbose",getDefaultVerbose());
	int const ranksplit = arginfo.getValue<int>("ranksplit",getDefaultRankSplit());
	int const rankstrip = arginfo.getValue<int>("rankstrip",getDefaultRankSplit());
	int const clipreinsert = arginfo.getValue<int>("clipreinsert",getDefaultClipReinsert());
	int const zztoname = arginfo.getValue<int>("zztoname",getDefaultZZToName());
	int const sanity = arginfo.getValue<int>("sanity",getDefaultSanity());
	uint64_t const mod = arginfo.getValue<int>("mod",getDefaultMod());
	uint64_t const bmod = libmaus2::math::nextTwoPow(mod);
	uint64_t const bmask = bmod-1;

	libmaus2::autoarray::AutoArray<char> Aread;

	::libmaus2::bambam::BamDecoder bamdec(std::cin,false);
	::libmaus2::bambam::BamHeader const & header = bamdec.getHeader();
	::libmaus2::bambam::BamHeader const & preheader = bampredec.getHeader();

	std::string const headertext(header.text);
	std::string const preheadertext(libmaus2::bambam::HeaderLine::removeSequenceLines(preheader.text));

	libmaus2::bambam::ProgramHeaderLineSet headerlines(headertext);
	libmaus2::bambam::ProgramHeaderLineSet preheaderlines(preheadertext);

	std::vector<libmaus2::bambam::HeaderLine> allheaderlines = libmaus2::bambam::HeaderLine::extractLines(headertext);

	std::string const lastid = preheaderlines.getLastIdInChain();

	std::stack < std::pair<uint64_t,std::string> > pgtodo;
	for ( uint64_t i = 0; i < headerlines.roots.size(); ++i )
		pgtodo.push(std::pair<uint64_t,std::string>(headerlines.roots[i],lastid));

	std::string upheadtext = preheadertext;
	while ( pgtodo.size() )
	{
		uint64_t const hid = pgtodo.top().first;
		std::string const PP = pgtodo.top().second;
		pgtodo.pop();
		libmaus2::bambam::HeaderLine const & line = headerlines.lines[hid];

		// ID, PP, PN, CL, VN
		std::string       ID = (line.M.find("ID") != line.M.end()) ? line.M.find("ID")->second : "";
		std::string const PN = (line.M.find("PN") != line.M.end()) ? line.M.find("PN")->second : "";
		std::string const CL = (line.M.find("CL") != line.M.end()) ? line.M.find("CL")->second : "";
		std::string const VN = (line.M.find("VN") != line.M.end()) ? line.M.find("VN")->second : "";

		upheadtext = ::libmaus2::bambam::ProgramHeaderLineSet::addProgramLineRef(
			upheadtext,
			ID,
			PN,
			CL,
			PP,
			VN
		);

		if ( headerlines.edges.find(hid) != headerlines.edges.end() )
		{
			std::vector<uint64_t> const & children = headerlines.edges.find(hid)->second;

			for ( uint64_t j = 0; j < children.size(); ++j )
				pgtodo.push(std::pair<uint64_t,std::string>(children[j],ID));
		}
	}

	/* copy SQ lines */
	std::ostringstream sqconcstr;
	sqconcstr << upheadtext;
	for ( uint64_t i = 0; i < allheaderlines.size(); ++i )
		if ( allheaderlines[i].type == "SQ" )
			sqconcstr << allheaderlines[i].line << "\n";
	upheadtext = sqconcstr.str();

	::libmaus2::bambam::BamHeader uphead(upheadtext);
	uphead.changeSortOrder("unknown");

	/*
	 * start index/md5 callbacks
	 */
	std::string const tmpfilenamebase = arginfo.getValue<std::string>("tmpfile",arginfo.getDefaultTmpFileName());
	std::string const tmpfileindex = tmpfilenamebase + "_index";
	::libmaus2::util::TempFileRemovalContainer::addTempFile(tmpfileindex);

	std::string md5filename;
	std::string indexfilename;

	std::vector< ::libmaus2::lz::BgzfDeflateOutputCallback * > cbs;
	::libmaus2::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Pmd5cb;
	if ( arginfo.getValue<unsigned int>("md5",getDefaultMD5()) )
	{
		if ( arginfo.hasArg("md5filename") &&  arginfo.getUnparsedValue("md5filename","") != "" )
			md5filename = arginfo.getUnparsedValue("md5filename","");
		else
			std::cerr << "[V] no filename for md5 given, not creating hash" << std::endl;

		if ( md5filename.size() )
		{
			::libmaus2::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Tmd5cb(new ::libmaus2::lz::BgzfDeflateOutputCallbackMD5);
			Pmd5cb = UNIQUE_PTR_MOVE(Tmd5cb);
			cbs.push_back(Pmd5cb.get());
		}
	}
	libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Pindex;
	if ( arginfo.getValue<unsigned int>("index",getDefaultIndex()) )
	{
		if ( arginfo.hasArg("indexfilename") &&  arginfo.getUnparsedValue("indexfilename","") != "" )
			indexfilename = arginfo.getUnparsedValue("indexfilename","");
		else
			std::cerr << "[V] no filename for index given, not creating index" << std::endl;

		if ( indexfilename.size() )
		{
			libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Tindex(new libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex(tmpfileindex));
			Pindex = UNIQUE_PTR_MOVE(Tindex);
			cbs.push_back(Pindex.get());
		}
	}
	std::vector< ::libmaus2::lz::BgzfDeflateOutputCallback * > * Pcbs = 0;
	if ( cbs.size() )
		Pcbs = &cbs;
	/*
	 * end md5/index callbacks
	 */

	::libmaus2::bambam::BamWriter::unique_ptr_type writer(new ::libmaus2::bambam::BamWriter(std::cout,uphead,level,Pcbs));

	::libmaus2::bambam::BamAlignment & algn = bamdec.getAlignment();
	::libmaus2::bambam::BamAlignment & prealgn = bampredec.getAlignment();
	int64_t curid = -1;

	libmaus2::autoarray::AutoArray< std::pair<uint8_t,uint8_t> > auxpre;
	libmaus2::autoarray::AutoArray< std::pair<uint8_t,uint8_t> > auxnew;

	libmaus2::bambam::BamAuxFilterVector auxfilter;

	// helpers for clipReinsert
	libmaus2::autoarray::AutoArray < std::pair<uint8_t,uint8_t> > auxtags;
	libmaus2::autoarray::AutoArray<libmaus2::bambam::cigar_operation> cigop;
	std::stack < libmaus2::bambam::cigar_operation > hardstack;
	libmaus2::bambam::BamAlignment::D_array_type Tcigar;
	libmaus2::bambam::BamAuxFilterVector bafv;
	libmaus2::bambam::BamAuxFilterVector auxfilterout;
	auxfilterout.set('q','s');
	auxfilterout.set('q','q');

	// helpers for zztoname
 	libmaus2::bambam::BamAuxFilterVector zzbafv;
 	zzbafv.set('z','z');

    	// tag filters for secondary/supplementary reads
	libmaus2::bambam::BamAuxFilterVector auxfiltersec;

	auxfiltersec.set('q','s');
	auxfiltersec.set('q','q');
	auxfiltersec.set('a','s');
	auxfiltersec.set('a','h');
	auxfiltersec.set('a','a');
	auxfiltersec.set('a','f');
	auxfiltersec.set('a','r');
	auxfiltersec.set('a','3');

	// loop over aligned BAM file
	while ( bamdec.readAlignment() )
	{
		if ( ranksplit )
			split12(algn);

		// extract rank
		char const * name = algn.getName();
		char const * u1 = name;
		bool ok = true;
		uint64_t rank = 0;
		while ( *u1 && *u1 != '_' )
		{
			rank *= 10;
			rank += (*u1-'0');
			ok = ok && isdigit(*u1);
			++u1;
		}

		// unable to find rank?	write out as is and continue
		if ( ! ok )
		{
			algn.serialise(writer->getStream());
			continue;
		}

		// loop over unaligned BAM file
		while ( curid != static_cast<int64_t>(rank) )
		{
			bool const a_ok = bampredec.readAlignment();

			if ( ! a_ok )
			{
				libmaus2::exception::LibMausException se;
				se.getStream() << "Found unexpected EOF on file " << prefilename << std::endl;
				se.finish();
				throw se;
			}
			assert ( a_ok );
			++curid;

			if ( verbose && (! (curid & bmask)) )
				std::cerr << "[V] " << (curid / bmod) << std::endl;
		}

		if ( verbose > 1 )
			std::cerr << "Merging:\n" << algn.formatAlignment(header) << "\n" << prealgn.formatAlignment(preheader) << std::endl;

		uint64_t pretagnum = prealgn.enumerateAuxTags(auxpre);
		uint64_t newtagnum = algn.enumerateAuxTags(auxnew);

		// do some sanity checking
		if ( sanity )
		{
		    	// first do a name check
			char const * prename = prealgn.getName();
			u1++; // put on the first letter of readname

			if ( verbose > 1 )
    	    	    	    	std::cerr << "Sanity: comparing " << name << " and " << prename << std::endl;

			if ( !is_suffix(prename, u1) ) // names do not match
			{
			    	libmaus2::exception::LibMausException se;
			    	se.getStream() << "Sanity check failed on read names, found " << name << " and " << prename << std::endl;
				se.finish();
				throw se;
			}

			// now the names match so try the flags

			if ( !(algn.isPaired() == prealgn.isPaired() &&
			     algn.isRead1() == prealgn.isRead1() &&
			     algn.isRead2() == prealgn.isRead2()) )
			{
			    	libmaus2::exception::LibMausException se;
				se.getStream() << "Sanity check failed on flags, " << std::endl
				    	       << "Aligned " << name << " paired " << algn.isPaired() << " first " << algn.isRead1() << " last " << algn.isRead2() << std::endl
			    	    	       << "Unaligned " << prename << " paired " << prealgn.isPaired() << " first " << prealgn.isRead1() << " last " << prealgn.isRead2() << std::endl;
				se.finish();
				throw se;
			}

			if ( verbose > 1 )
			    std::cerr << "Sanity check on flags: " << std::endl
				    	       << "Aligned " << name << " paired " << algn.isPaired() << " first " << algn.isRead1() << " last " << algn.isRead2() << std::endl
			    	    	       << "Unaligned " << prename << " paired " << prealgn.isPaired() << " first " << prealgn.isRead1() << " last " << prealgn.isRead2() << std::endl;


		}

		std::sort(auxpre.begin(),auxpre.begin()+pretagnum);
		std::sort(auxnew.begin(),auxnew.begin()+newtagnum);

		if ( verbose > 1 )
			std::cerr << "pretagnum=" << pretagnum << " newtagnum=" << newtagnum << std::endl;

		std::pair<uint8_t,uint8_t> * prec = auxpre.begin();
		std::pair<uint8_t,uint8_t> * pree = prec + pretagnum;
		std::pair<uint8_t,uint8_t> * preo = prec;

		std::pair<uint8_t,uint8_t> * newc = auxnew.begin();
		std::pair<uint8_t,uint8_t> * newe = newc + newtagnum;
		std::pair<uint8_t,uint8_t> * newo = newc;

		while ( prec != pree && newc != newe )
		{
			// pre which is not in new
			if ( *prec < *newc )
			{
				*(preo++) = *(prec++);
			}
			// tag in both, drop pre
			else if ( *prec == *newc )
			{
				*(newo++) = *(newc++);
				prec++;
			}
			// new not in pre
			else
			{
				*(newo++) = *(newc++);
			}
		}

		while ( prec != pree )
			*(preo++) = *(prec++);
		while ( newc != newe )
			*(newo++) = *(newc++);

		pretagnum = preo-auxpre.begin();
		newtagnum = newo-auxnew.begin();

		for ( uint64_t i = 0; i < pretagnum; ++i )
			auxfilter.set(auxpre[i].first,auxpre[i].second);

		algn.copyAuxTags(prealgn, auxfilter);

		for ( uint64_t i = 0; i < pretagnum; ++i )
			auxfilter.clear(auxpre[i].first,auxpre[i].second);

		if ( verbose > 1 )
		{
			std::cerr << "pretagnum=" << pretagnum << " newtagnum=" << newtagnum << std::endl;
			std::cerr << "result: " << algn.formatAlignment(header) << std::endl;
		}


		if ( algn.isSecondary() || algn.isSupplementary() )
		{
		    	// adding adapter clip data to secondary/supplementary reads
			// can lead to incorrect clip reinserts so remove these tags

			algn.filterOutAux(auxfiltersec);
		}


		// copy QC fail flag from original file to aligner output
		if ( prealgn.isQCFail() )
			algn.putFlags( algn.getFlags() | libmaus2::bambam::BamFlagBase::LIBMAUS2_BAMBAM_FQCFAIL );

		if ( rankstrip )
			strip12(algn);

		if ( clipreinsert )
			clipReinsert(algn,auxtags,bafv,cigop,Tcigar,hardstack,auxfilterout);

		if ( zztoname )
			zzToRank(algn,zzbafv);

		algn.serialise(writer->getStream());
	}

	writer.reset();

	if ( Pmd5cb )
	{
		Pmd5cb->saveDigestAsFile(md5filename);
	}
	if ( Pindex )
	{
		Pindex->flush(std::string(indexfilename));
	}

	return EXIT_SUCCESS;
}
Esempio n. 21
0
int bamfilter(libmaus::util::ArgInfo const & arginfo)
{
	uint64_t const minmapped = arginfo.getValue<uint64_t>("minmapped",getDefaultMinMapped());
	uint64_t const maxmapped = arginfo.getValue<uint64_t>("maxmapped",getDefaultMaxMapped());
	uint64_t const minlen = arginfo.getValue<uint64_t>("minlen",getDefaultMinLen());
	int const level = libmaus::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(arginfo.getValue<int>("level",getDefaultLevel()));
	
	::libmaus::bambam::BamDecoder BD(std::cin);
	::libmaus::bambam::BamHeader const & bamheader = BD.getHeader();
	::libmaus::bambam::BamAlignment & alignment = BD.getAlignment();

	/*
	 * start index/md5 callbacks
	 */
	std::string const tmpfilenamebase = arginfo.getValue<std::string>("tmpfile",arginfo.getDefaultTmpFileName());
	std::string const tmpfileindex = tmpfilenamebase + "_index";
	::libmaus::util::TempFileRemovalContainer::addTempFile(tmpfileindex);

	std::string md5filename;
	std::string indexfilename;

	std::vector< ::libmaus::lz::BgzfDeflateOutputCallback * > cbs;
	::libmaus::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Pmd5cb;
	if ( arginfo.getValue<unsigned int>("md5",getDefaultMD5()) )
	{
		if ( arginfo.hasArg("md5filename") &&  arginfo.getUnparsedValue("md5filename","") != "" )
			md5filename = arginfo.getUnparsedValue("md5filename","");
		else
			std::cerr << "[V] no filename for md5 given, not creating hash" << std::endl;

		if ( md5filename.size() )
		{
			::libmaus::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Tmd5cb(new ::libmaus::lz::BgzfDeflateOutputCallbackMD5);
			Pmd5cb = UNIQUE_PTR_MOVE(Tmd5cb);
			cbs.push_back(Pmd5cb.get());
		}
	}
	libmaus::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Pindex;
	if ( arginfo.getValue<unsigned int>("index",getDefaultIndex()) )
	{
		if ( arginfo.hasArg("indexfilename") &&  arginfo.getUnparsedValue("indexfilename","") != "" )
			indexfilename = arginfo.getUnparsedValue("indexfilename","");
		else
			std::cerr << "[V] no filename for index given, not creating index" << std::endl;

		if ( indexfilename.size() )
		{
			libmaus::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Tindex(new libmaus::bambam::BgzfDeflateOutputCallbackBamIndex(tmpfileindex));
			Pindex = UNIQUE_PTR_MOVE(Tindex);
			cbs.push_back(Pindex.get());
		}
	}
	std::vector< ::libmaus::lz::BgzfDeflateOutputCallback * > * Pcbs = 0;
	if ( cbs.size() )
		Pcbs = &cbs;
	/*
	 * end md5/index callbacks
	 */

	::libmaus::bambam::BamHeader::unique_ptr_type uphead(libmaus::bambam::BamHeaderUpdate::updateHeader(arginfo,bamheader,"bamfilter",std::string(PACKAGE_VERSION)));
	::libmaus::bambam::BamWriter::unique_ptr_type writer(new ::libmaus::bambam::BamWriter(std::cout,*uphead,level,Pcbs));
	
	while ( BD.readAlignment() )
	{
		bool const a_1_mapped = !(alignment.getFlags() & ::libmaus::bambam::BamFlagBase::LIBMAUS_BAMBAM_FUNMAP);
		bool const a_2_mapped = !(alignment.getFlags() & ::libmaus::bambam::BamFlagBase::LIBMAUS_BAMBAM_FMUNMAP);
		bool const proper     =  (alignment.getFlags() & ::libmaus::bambam::BamFlagBase::LIBMAUS_BAMBAM_FPROPER_PAIR);

		uint64_t const nummapped = (a_1_mapped?1:0)+(a_2_mapped?1:0)+(proper?1:0);

		if ( 
			nummapped >= minmapped && 
			nummapped <= maxmapped && 
			alignment.getLseq() >= static_cast<int64_t>(minlen)
		)
			alignment.serialise(writer->getStream());
	}	

	writer.reset();

	if ( Pmd5cb )
	{
		Pmd5cb->saveDigestAsFile(md5filename);
	}
	if ( Pindex )
	{
		Pindex->flush(std::string(indexfilename));
	}
	
	return EXIT_SUCCESS;
}
Esempio n. 22
0
uint64_t bamrecompress(libmaus2::util::ArgInfo const & arginfo)
{
	int const level = libmaus2::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(arginfo.getValue<int>("level",getDefaultLevel()));
	int const verbose = arginfo.getValue<int>("verbose",getDefaultVerbose());
	int const numthreads = std::max(1,arginfo.getValue<int>("numthreads",getDefaultNumThreads()));

	/*
	 * start index/md5 callbacks
	 */
	std::string const tmpfilenamebase = arginfo.getValue<std::string>("tmpfile",arginfo.getDefaultTmpFileName());
	std::string const tmpfileindex = tmpfilenamebase + "_index";
	::libmaus2::util::TempFileRemovalContainer::addTempFile(tmpfileindex);

	std::string md5filename;
	std::string indexfilename;

	std::vector< ::libmaus2::lz::BgzfDeflateOutputCallback * > cbs;
	::libmaus2::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Pmd5cb;
	if ( arginfo.getValue<unsigned int>("md5",getDefaultMD5()) )
	{
		if ( arginfo.hasArg("md5filename") &&  arginfo.getUnparsedValue("md5filename","") != "" )
			md5filename = arginfo.getUnparsedValue("md5filename","");
		else
			std::cerr << "[V] no filename for md5 given, not creating hash" << std::endl;

		if ( md5filename.size() )
		{
			::libmaus2::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Tmd5cb(new ::libmaus2::lz::BgzfDeflateOutputCallbackMD5);
			Pmd5cb = UNIQUE_PTR_MOVE(Tmd5cb);
			cbs.push_back(Pmd5cb.get());
		}
	}
	libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Pindex;
	if ( arginfo.getValue<unsigned int>("index",getDefaultIndex()) )
	{
		if ( arginfo.hasArg("indexfilename") &&  arginfo.getUnparsedValue("indexfilename","") != "" )
			indexfilename = arginfo.getUnparsedValue("indexfilename","");
		else
			std::cerr << "[V] no filename for index given, not creating index" << std::endl;

		if ( indexfilename.size() )
		{
			libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Tindex(new libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex(tmpfileindex));
			Pindex = UNIQUE_PTR_MOVE(Tindex);
			cbs.push_back(Pindex.get());
		}
	}
	/*
	 * end md5/index callbacks
	 */

	libmaus2::lz::BgzfInflateDeflateParallel::unique_ptr_type BIDP(new libmaus2::lz::BgzfInflateDeflateParallel(std::cin,std::cout,level,numthreads,4*numthreads));

	for ( uint64_t i = 0; i < cbs.size(); ++i )
		BIDP->registerBlockOutputCallback(cbs[i]);

	libmaus2::autoarray::AutoArray<char> B(64*1024,false);
	int r;
	uint64_t t = 0;
	uint64_t last = std::numeric_limits<uint64_t>::max();
	uint64_t lcnt = 0;
	uint64_t const mod = 64*1024*1024;
	libmaus2::timing::RealTimeClock rtc; rtc.start();
	libmaus2::timing::RealTimeClock lrtc; lrtc.start();

	while ( (r = BIDP->read(B.begin(),B.size())) )
	{
		BIDP->write(B.begin(),r);

		lcnt += r;
		t += r;

		if ( t/mod != last/mod )
		{
			if ( verbose )
			{
				if ( isatty(STDERR_FILENO) )
					std::cerr
						<< "\r" << std::string(60,' ') << "\r";

				std::cerr
						<< rtc.formatTime(rtc.getElapsedSeconds()) << " " << t/(1024*1024) << "MB, " << (lcnt/lrtc.getElapsedSeconds())/(1024.0*1024.0) << "MB/s";

				if ( isatty(STDERR_FILENO) )
					std::cerr << std::flush;
				else
					std::cerr << std::endl;
			}

			lrtc.start();
			last = t;
			lcnt = 0;
		}
	}

	if ( verbose )
	{
		if ( isatty(STDERR_FILENO) )
			std::cerr
				<< "\r" << std::string(60,' ') << "\r";

		std::cerr
				<< rtc.formatTime(rtc.getElapsedSeconds()) << " " << t/(1024*1024) << "MB, " << (t/rtc.getElapsedSeconds())/(1024.0*1024.0) << "MB/s";

		std::cerr << std::endl;
	}

	BIDP.reset();

	if ( Pmd5cb )
	{
		Pmd5cb->saveDigestAsFile(md5filename);
	}
	if ( Pindex )
	{
		Pindex->flush(std::string(indexfilename));
	}

	return 0;
}
Esempio n. 23
0
int bamclipreinsert(::libmaus2::util::ArgInfo const & arginfo)
{
	if ( isatty(STDIN_FILENO) )
	{
		::libmaus2::exception::LibMausException se;
		se.getStream() << "Refusing to read binary data from terminal, please redirect standard input to pipe or file." << std::endl;
		se.finish();
		throw se;
	}

	if ( isatty(STDOUT_FILENO) )
	{
		::libmaus2::exception::LibMausException se;
		se.getStream() << "Refusing write binary data to terminal, please redirect standard output to pipe or file." << std::endl;
		se.finish();
		throw se;
	}

	int const level = libmaus2::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(arginfo.getValue<int>("level",getDefaultLevel()));
	int const verbose = arginfo.getValue<int>("verbose",getDefaultVerbose());

	::libmaus2::bambam::BamDecoder dec(std::cin,false);
	::libmaus2::bambam::BamHeader const & header = dec.getHeader();

	std::string const headertext(header.text);

	// add PG line to header
	std::string const upheadtext = ::libmaus2::bambam::ProgramHeaderLineSet::addProgramLine(
		headertext,
		"bamclipreinsert", // ID
		"bamclipreinsert", // PN
		arginfo.commandline, // CL
		::libmaus2::bambam::ProgramHeaderLineSet(headertext).getLastIdInChain(), // PP
		std::string(PACKAGE_VERSION) // VN
	);

	// construct new header
	libmaus2::bambam::BamHeader const uphead(upheadtext);

	/*
	 * start index/md5 callbacks
	 */
	std::string const tmpfilenamebase = arginfo.getValue<std::string>("tmpfile",arginfo.getDefaultTmpFileName());
	std::string const tmpfileindex = tmpfilenamebase + "_index";
	::libmaus2::util::TempFileRemovalContainer::addTempFile(tmpfileindex);

	std::string md5filename;
	std::string indexfilename;

	std::vector< ::libmaus2::lz::BgzfDeflateOutputCallback * > cbs;
	::libmaus2::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Pmd5cb;
	if ( arginfo.getValue<unsigned int>("md5",getDefaultMD5()) )
	{
		if ( arginfo.hasArg("md5filename") &&  arginfo.getUnparsedValue("md5filename","") != "" )
			md5filename = arginfo.getUnparsedValue("md5filename","");
		else
			std::cerr << "[V] no filename for md5 given, not creating hash" << std::endl;

		if ( md5filename.size() )
		{
			::libmaus2::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Tmd5cb(new ::libmaus2::lz::BgzfDeflateOutputCallbackMD5);
			Pmd5cb = UNIQUE_PTR_MOVE(Tmd5cb);
			cbs.push_back(Pmd5cb.get());
		}
	}
	libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Pindex;
	if ( arginfo.getValue<unsigned int>("index",getDefaultIndex()) )
	{
		if ( arginfo.hasArg("indexfilename") &&  arginfo.getUnparsedValue("indexfilename","") != "" )
			indexfilename = arginfo.getUnparsedValue("indexfilename","");
		else
			std::cerr << "[V] no filename for index given, not creating index" << std::endl;

		if ( indexfilename.size() )
		{
			libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Tindex(new libmaus2::bambam::BgzfDeflateOutputCallbackBamIndex(tmpfileindex));
			Pindex = UNIQUE_PTR_MOVE(Tindex);
			cbs.push_back(Pindex.get());
		}
	}
	std::vector< ::libmaus2::lz::BgzfDeflateOutputCallback * > * Pcbs = 0;
	if ( cbs.size() )
		Pcbs = &cbs;
	/*
	 * end md5/index callbacks
	 */

	::libmaus2::bambam::BamWriter::unique_ptr_type writer(new ::libmaus2::bambam::BamWriter(std::cout,uphead,level,Pcbs));
	libmaus2::bambam::BamAuxFilterVector bafv;
 	// bafv.set('z','z');
 	// std::vector<uint8_t> R(8);
 	// std::string const zz("zz");

	libmaus2::bambam::BamAlignment & algn = dec.getAlignment();
	uint64_t c = 0;

	libmaus2::autoarray::AutoArray < std::pair<uint8_t,uint8_t> > auxtags;
	libmaus2::autoarray::AutoArray<libmaus2::bambam::cigar_operation> cigop;
	std::stack < libmaus2::bambam::cigar_operation > hardstack;
	libmaus2::bambam::BamAlignment::D_array_type Tcigar;
	libmaus2::bambam::BamAuxFilterVector auxfilterout;
	auxfilterout.set('q','s');
	auxfilterout.set('q','q');

	while ( dec.readAlignment() )
	{
		// reinsert clipped parts and attach soft clipping cigar operations as needed
		clipReinsert(algn,auxtags,bafv,cigop,Tcigar,hardstack,auxfilterout);

		algn.serialise(writer->getStream());

		++c;

		if ( verbose && (c & (1024*1024-1)) == 0 )
 			std::cerr << "[V] " << c/(1024*1024) << std::endl;
	}

	writer.reset();

	if ( Pmd5cb )
	{
		Pmd5cb->saveDigestAsFile(md5filename);
	}
	if ( Pindex )
	{
		Pindex->flush(std::string(indexfilename));
	}

	return EXIT_SUCCESS;
}
Esempio n. 24
0
int main(int argc, char * argv[])
{
    try
    {
        ::libmaus::util::ArgInfo const arginfo(argc,argv);

        for ( uint64_t i = 0; i < arginfo.restargs.size(); ++i )
            if (
                arginfo.restargs[i] == "-v"
                ||
                arginfo.restargs[i] == "--version"
            )
            {
                std::cerr << ::biobambam::Licensing::license();
                return EXIT_SUCCESS;
            }
            else if (
                arginfo.restargs[i] == "-h"
                ||
                arginfo.restargs[i] == "--help"
            )
            {
                std::cerr << ::biobambam::Licensing::license();
                std::cerr << std::endl;
                std::cerr << "Key=Value pairs:" << std::endl;
                std::cerr << std::endl;

                std::vector< std::pair<std::string,std::string> > V;

                V.push_back ( std::pair<std::string,std::string> ( "div=<["+::biobambam::Licensing::formatNumber(getDefaultDiv())+"]>", "divisor" ) );
                V.push_back ( std::pair<std::string,std::string> ( "level=<["+::biobambam::Licensing::formatNumber(getDefaultLevel())+"]>", libmaus::bambam::BamBlockWriterBaseFactory::getBamOutputLevelHelpText() ) );
                V.push_back ( std::pair<std::string,std::string> ( "verbose=<["+::biobambam::Licensing::formatNumber(getDefaultVerbose())+"]>", "print progress report" ) );
                V.push_back ( std::pair<std::string,std::string> ( "prefix=<["+getDefaultFilePrefix(arginfo)+"]>", "default output file prefix" ) );

                ::biobambam::Licensing::printMap(std::cerr,V);

                std::cerr << std::endl;
                return EXIT_SUCCESS;
            }

        return bamsplitmod(arginfo);
    }
    catch(std::exception const & ex)
    {
        std::cerr << ex.what() << std::endl;
        return EXIT_FAILURE;
    }
}
Esempio n. 25
0
void normalisefastaBgzf(libmaus::util::ArgInfo const & arginfo, std::ostream & out)
{
	libmaus::fastx::StreamFastAReaderWrapper in(std::cin);
	libmaus::fastx::StreamFastAReaderWrapper::pattern_type pattern;
	int const level = libmaus::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(arginfo.getValue("level",getDefaultLevel()));
	std::string const indexfn = arginfo.getUnparsedValue("index","");

	::libmaus::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(level);

	libmaus::lz::BgzfDeflate<std::ostream> defl(out,level,false /* full flush */);
	uint64_t const inbufsize = defl.getInputBufferSize();
	uint64_t zoffset = 0;
	uint64_t ioffset = 0;
	std::vector<libmaus::fastx::BgzfFastAIndexEntry> index;
	std::ostringstream indexstr;
	
	ioffset += libmaus::util::NumberSerialisation::serialiseNumber(indexstr,inbufsize);
	uint64_t patid = 0;
	
	while ( in.getNextPatternUnlocked(pattern) )
	{
		std::string const name = pattern.getStringId();
		std::string const shortname = stripName(name);
		std::string const & spat = pattern.spattern;
		char const * cpat = spat.c_str();
		uint64_t const patlen = spat.size();
		uint64_t const numblocks = (patlen + inbufsize - 1)/inbufsize;

		index.push_back(libmaus::fastx::BgzfFastAIndexEntry(shortname,patid++,ioffset));
		
		ioffset += libmaus::util::StringSerialisation::serialiseString(indexstr,name);
		ioffset += libmaus::util::StringSerialisation::serialiseString(indexstr,shortname);
		ioffset += libmaus::util::NumberSerialisation::serialiseNumber(indexstr,patlen);
		ioffset += libmaus::util::NumberSerialisation::serialiseNumber(indexstr,zoffset);
		ioffset += libmaus::util::NumberSerialisation::serialiseNumber(indexstr,numblocks);
		
		std::ostringstream nameostr;
		nameostr << '>' << name << '\n';
		std::string const nameser = nameostr.str();
				
		std::pair<uint64_t,uint64_t> const P0 = defl.writeSyncedCount(nameser.c_str(),nameser.size());
		zoffset += P0.second;
		
		uint64_t o = 0;
		while ( o != patlen )
		{
			assert ( o % inbufsize == 0 );
			uint64_t const towrite = std::min(patlen-o,inbufsize);
			std::pair<uint64_t,uint64_t> const P1 = defl.writeSyncedCount(cpat,towrite);
			
			ioffset += libmaus::util::NumberSerialisation::serialiseNumber(indexstr,zoffset);

			zoffset += P1.second;
			o += towrite;
			cpat += towrite;
		}		

		ioffset += libmaus::util::NumberSerialisation::serialiseNumber(indexstr,zoffset);

		std::pair<uint64_t,uint64_t> const Pn = defl.writeSyncedCount("\n",1);
		zoffset += Pn.second;
	}

	defl.flush();
	out << std::flush;
	
	uint64_t const imetaoffset = ioffset;

	ioffset += libmaus::util::NumberSerialisation::serialiseNumber(indexstr,index.size());
	for ( uint64_t i = 0; i < index.size(); ++i )
		ioffset += libmaus::util::NumberSerialisation::serialiseNumber(indexstr,index[i].ioffset);
	
	libmaus::util::NumberSerialisation::serialiseNumber(indexstr,imetaoffset);

	if ( indexfn.size() )
	{
		std::string const & sindex = indexstr.str();
		libmaus::aio::CheckedOutputStream indexCOS(indexfn);
		indexCOS.write(sindex.c_str(),sindex.size());
		indexCOS.flush();
		indexCOS.close();	
	}
}
Esempio n. 26
0
int bamsplitmod(libmaus::util::ArgInfo const & arginfo)
{
    if ( isatty(STDIN_FILENO) )
    {
        ::libmaus::exception::LibMausException se;
        se.getStream() << "Refusing read binary data from terminal, please redirect standard input to pipe or file." << std::endl;
        se.finish();
        throw se;
    }

    int const level = libmaus::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(arginfo.getValue<int>("level",getDefaultLevel()));
    int const verbose = arginfo.getValue<int>("verbose",getDefaultVerbose());
    uint64_t const div = arginfo.getValue<int>("div",getDefaultDiv());
    std::string const prefix = arginfo.getUnparsedValue("prefix",getDefaultFilePrefix(arginfo));

    if ( ! div )
    {
        ::libmaus::exception::LibMausException se;
        se.getStream() << "div cannot be 0." << std::endl;
        se.finish();
        throw se;
    }

    libmaus::bambam::BamDecoder bamdec(std::cin);
    libmaus::bambam::BamAlignment const & algn = bamdec.getAlignment();
    libmaus::bambam::BamHeader const & header = bamdec.getHeader();
    ::libmaus::bambam::BamHeader::unique_ptr_type uphead(updateHeader(arginfo,header));

    libmaus::autoarray::AutoArray<libmaus::aio::CheckedOutputStream::unique_ptr_type> COS(div);
    libmaus::autoarray::AutoArray<libmaus::bambam::BamWriter::unique_ptr_type> writers(div);
    std::vector < std::string > filenames;
    for ( uint64_t i = 0; i < div; ++i )
    {
        std::ostringstream ostr;
        ostr << prefix << "_" << std::setw(6) << std::setfill('0') << i << std::setw(0) << ".bam";

        libmaus::aio::CheckedOutputStream::unique_ptr_type tCOS(new libmaus::aio::CheckedOutputStream(ostr.str()));
        COS[i] = UNIQUE_PTR_MOVE(tCOS);
        libmaus::bambam::BamWriter::unique_ptr_type twriter(new libmaus::bambam::BamWriter(*COS[i],*uphead,level));
        writers[i] = UNIQUE_PTR_MOVE(twriter);
    }

    uint64_t c = 0;
    if ( verbose )
    {
        while ( bamdec.readAlignment() )
        {
            algn.serialise ( writers [ (c++) % div ] -> getStream() );

            if ( ((c) & ((1ull<<20)-1)) == 0 )
                std::cerr << "[V] " << c << std::endl;
        }
        std::cerr << "[V] " << c << std::endl;
    }
    else
    {
        while ( bamdec.readAlignment() )
            algn.serialise ( writers [ (c++) % div ] -> getStream() );
    }

    for ( uint64_t i = 0; i < div; ++i )
    {
        writers[i].reset();
        COS[i]->flush();
        COS[i].reset();
    }

    return EXIT_SUCCESS;
}
Esempio n. 27
0
int bamfixmatecoordinates(::libmaus::util::ArgInfo const & arginfo)
{
	::libmaus::util::TempFileRemovalContainer::setup();
	::libmaus::timing::RealTimeClock rtc; rtc.start();
	
	bool const verbose = arginfo.getValue<unsigned int>("verbose",getDefaultVerbose());
	unsigned int const colhashbits = arginfo.getValue<unsigned int>("colhashbits",getDefaultColHashBits());
	unsigned int const collistsize = arginfo.getValue<unsigned int>("collistsize",getDefaultColListSize());
	int const level = arginfo.getValue<int>("level",getDefaultLevel());
	std::string const tmpfilenamebase = arginfo.getValue<std::string>("tmpfile",arginfo.getDefaultTmpFileName());
	
	switch ( level )
	{
		case Z_NO_COMPRESSION:
		case Z_BEST_SPEED:
		case Z_BEST_COMPRESSION:
		case Z_DEFAULT_COMPRESSION:
			break;
		default:
		{
			::libmaus::exception::LibMausException se;
			se.getStream()
				<< "Unknown compression level, please use"
				<< " level=" << Z_DEFAULT_COMPRESSION << " (default) or"
				<< " level=" << Z_BEST_SPEED << " (fast) or"
				<< " level=" << Z_BEST_COMPRESSION << " (best) or"
				<< " level=" << Z_NO_COMPRESSION << " (no compression)" << std::endl;
			se.finish();
			throw se;
		}
			break;
	}


	std::string const tmpfilename = tmpfilenamebase + "_bamcollate";
	::libmaus::util::TempFileRemovalContainer::addTempFile(tmpfilename);
	
	::libmaus::bambam::CollatingBamDecoder CBD(std::cin,tmpfilename,false /* put rank */,colhashbits/*hash bits*/,collistsize/*size of output list*/);
	::libmaus::bambam::BamFormatAuxiliary auxdata;
	::libmaus::bambam::BamHeader const & bamheader = CBD.getHeader();
	
	// add PG line to header
	std::string const upheadtext = ::libmaus::bambam::ProgramHeaderLineSet::addProgramLine(
		bamheader.text,
		"bamfixmatecoordinates", // ID
		"bamfixmatecoordinates", // PN
		arginfo.commandline, // CL
		::libmaus::bambam::ProgramHeaderLineSet(bamheader.text).getLastIdInChain(), // PP
		std::string(PACKAGE_VERSION) // VN			
	);
	// construct new header
	::libmaus::bambam::BamHeader uphead(upheadtext);
	
	if ( uphead.getSortOrder() != "queryname" )
		uphead.changeSortOrder("unknown");

	/*
	 * start index/md5 callbacks
	 */
	std::string const tmpfileindex = tmpfilenamebase + "_index";
	::libmaus::util::TempFileRemovalContainer::addTempFile(tmpfileindex);

	std::string md5filename;
	std::string indexfilename;

	std::vector< ::libmaus::lz::BgzfDeflateOutputCallback * > cbs;
	::libmaus::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Pmd5cb;
	if ( arginfo.getValue<unsigned int>("md5",getDefaultMD5()) )
	{
		if ( arginfo.hasArg("md5filename") &&  arginfo.getUnparsedValue("md5filename","") != "" )
			md5filename = arginfo.getUnparsedValue("md5filename","");
		else
			std::cerr << "[V] no filename for md5 given, not creating hash" << std::endl;

		if ( md5filename.size() )
		{
			::libmaus::lz::BgzfDeflateOutputCallbackMD5::unique_ptr_type Tmd5cb(new ::libmaus::lz::BgzfDeflateOutputCallbackMD5);
			Pmd5cb = UNIQUE_PTR_MOVE(Tmd5cb);
			cbs.push_back(Pmd5cb.get());
		}
	}
	libmaus::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Pindex;
	if ( arginfo.getValue<unsigned int>("index",getDefaultIndex()) )
	{
		if ( arginfo.hasArg("indexfilename") &&  arginfo.getUnparsedValue("indexfilename","") != "" )
			indexfilename = arginfo.getUnparsedValue("indexfilename","");
		else
			std::cerr << "[V] no filename for index given, not creating index" << std::endl;

		if ( indexfilename.size() )
		{
			libmaus::bambam::BgzfDeflateOutputCallbackBamIndex::unique_ptr_type Tindex(new libmaus::bambam::BgzfDeflateOutputCallbackBamIndex(tmpfileindex));
			Pindex = UNIQUE_PTR_MOVE(Tindex);
			cbs.push_back(Pindex.get());
		}
	}
	std::vector< ::libmaus::lz::BgzfDeflateOutputCallback * > * Pcbs = 0;
	if ( cbs.size() )
		Pcbs = &cbs;
	/*
	 * end md5/index callbacks
	 */
	
	// setup bam writer
	::libmaus::bambam::BamWriter::unique_ptr_type writer(new ::libmaus::bambam::BamWriter(std::cout,uphead,level,Pcbs));
	
	#if 0
	::libmaus::bambam::ProgramHeaderLineSet PHLS(bamheader.text);
	std::cerr << "Last id in PG chain: " << PHLS.getLastIdInChain() << std::endl;
	#endif

	// std::cout << bamheader.text;

	typedef ::libmaus::bambam::CollatingBamDecoder::alignment_ptr_type alignment_ptr_type;
	std::pair<alignment_ptr_type,alignment_ptr_type> P;
	uint64_t const mod = 1024*1024;
	uint64_t proc = 0;
	uint64_t lastproc = 0;
	uint64_t paircnt = 0;
	
	while ( CBD.tryPair(P) )
	{
		uint64_t const mapcnt = getMapCnt(P.first) + getMapCnt(P.second);
		
		if ( mapcnt == 1 )
		{
			int32_t refid = -1;
			int32_t pos = -1;
			
			if ( P.first )
			{
				refid = P.first->getRefID();
				pos = P.first->getPos();
			}
			else
			{
				assert ( P.second );

				refid = P.second->getRefID();
				pos = P.second->getPos();
			}
			
			P.first->putRefId(refid);
			P.first->putPos(pos);
			P.first->putNextRefId(refid);
			P.first->putNextPos(pos);
			P.second->putRefId(refid);
			P.second->putPos(pos);
			P.second->putNextRefId(refid);
			P.second->putNextPos(pos);
		}
		
		if ( P.first )
		{
			P.first->serialise(writer->getStream());
			++proc;
		}
		if ( P.second )
		{
			P.second->serialise(writer->getStream());
			++proc;
		}
		if ( P.first && P.second )
		{
			paircnt++;
		}
		
		if ( verbose && (proc/mod != lastproc/mod) )
		{
			std::cerr 
				<< "Processed " << proc << " fragments, " << paircnt << " pairs, " 
				<< proc/rtc.getElapsedSeconds() << " al/s"
				<< std::endl;
			lastproc = proc;
		}
	}		

	if ( verbose )
		std::cerr 	
			<< "Processed " << proc << " fragments, " << paircnt << " pairs, " 
			<< proc/rtc.getElapsedSeconds() << " al/s"
			<< std::endl;

	writer.reset();

	if ( Pmd5cb )
	{
		Pmd5cb->saveDigestAsFile(md5filename);
	}
	if ( Pindex )
	{
		Pindex->flush(std::string(indexfilename));
	}

	return EXIT_SUCCESS;
}
Esempio n. 28
0
int bamsplit(libmaus2::util::ArgInfo const & arginfo)
{
	if ( isatty(STDIN_FILENO) )
	{
		::libmaus2::exception::LibMausException se;
		se.getStream() << "Refusing read binary data from terminal, please redirect standard input to pipe or file." << std::endl;
		se.finish();
		throw se;
	}

	int const level = libmaus2::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(arginfo.getValue<int>("level",getDefaultLevel()));
	int const verbose = arginfo.getValue<int>("verbose",getDefaultVerbose());
	uint64_t const n = arginfo.getValue<int>("n",getDefaultN());
	std::string const prefix = arginfo.getUnparsedValue("prefix",getDefaultFilePrefix(arginfo));

	libmaus2::bambam::BamDecoder bamdec(std::cin);
	libmaus2::bambam::BamAlignment const & algn = bamdec.getAlignment();
	libmaus2::bambam::BamHeader const & header = bamdec.getHeader();
	::libmaus2::bambam::BamHeader::unique_ptr_type uphead(updateHeader(arginfo,header));

	libmaus2::aio::OutputStreamInstance::unique_ptr_type COS;
	libmaus2::bambam::BamWriter::unique_ptr_type writer;

	uint64_t c = 0;
	uint64_t f = 0;
	while ( bamdec.readAlignment() )
	{
		if ( c++ % n == 0 )
		{
			writer.reset();
			if ( COS )
				COS->flush();
			COS.reset();

			std::ostringstream fnostr;
			fnostr << prefix << "_" << std::setw(6) << std::setfill('0') << f++ << std::setw(0) << ".bam";
			std::string const fn = fnostr.str();

			libmaus2::aio::OutputStreamInstance::unique_ptr_type tCOS(new libmaus2::aio::OutputStreamInstance(fn));
			COS = UNIQUE_PTR_MOVE(tCOS);

			libmaus2::bambam::BamWriter::unique_ptr_type twriter(new libmaus2::bambam::BamWriter(*COS,*uphead,level));
			writer = UNIQUE_PTR_MOVE(twriter);

			if ( verbose )
				std::cerr << "[V] opened file " << fn << std::endl;
		}

		algn.serialise(writer->getStream());
	}

	writer.reset();
	if ( COS )
		COS->flush();
	COS.reset();

	return EXIT_SUCCESS;
}