// ***************************************************************************
// bamtools_convert.cpp (c) 2010 Derek Barnett, Erik Garrison
// Marth Lab, Department of Biology, Boston College
-// All rights reserved.
// ---------------------------------------------------------------------------
-// Last modified: 16 September 2010
+// Last modified: 11 November 2012
// ---------------------------------------------------------------------------
// Converts between BAM and a number of other formats
// ***************************************************************************
+#include "bamtools_convert.h"
+
+#include <api/BamConstants.h>
+#include <api/BamMultiReader.h>
+#include <utils/bamtools_fasta.h>
+#include <utils/bamtools_options.h>
+#include <utils/bamtools_pileup_engine.h>
+#include <utils/bamtools_utilities.h>
+using namespace BamTools;
+
#include <fstream>
#include <iostream>
#include <sstream>
#include <string>
#include <vector>
-#include "bamtools_convert.h"
-#include "bamtools_fasta.h"
-#include "bamtools_options.h"
-#include "bamtools_pileup_engine.h"
-#include "bamtools_utilities.h"
-#include "BGZF.h"
-#include "BamMultiReader.h"
using namespace std;
-using namespace BamTools;
namespace BamTools {
- // ---------------------------------------------
- // ConvertTool constants
-
- // supported conversion format command-line names
- static const string FORMAT_BED = "bed";
- static const string FORMAT_BEDGRAPH = "bedgraph";
- static const string FORMAT_FASTA = "fasta";
- static const string FORMAT_FASTQ = "fastq";
- static const string FORMAT_JSON = "json";
- static const string FORMAT_SAM = "sam";
- static const string FORMAT_PILEUP = "pileup";
- static const string FORMAT_WIGGLE = "wig";
-
- // other constants
- static const unsigned int FASTA_LINE_MAX = 50;
-
- // ---------------------------------------------
- // ConvertPileupFormatVisitor declaration
-
- class ConvertPileupFormatVisitor : public PileupVisitor {
-
- // ctor & dtor
- public:
- ConvertPileupFormatVisitor(const RefVector& references,
- const string& fastaFilename,
- const bool isPrintingMapQualities,
- ostream* out);
- ~ConvertPileupFormatVisitor(void);
-
- // PileupVisitor interface implementation
- public:
-// void Visit(const int& refId, const int& position, const vector<BamAlignment>& alignments);
-
- void Visit(const PileupPosition& pileupData);
-
- // data members
- private:
- Fasta m_fasta;
- bool m_hasFasta;
- bool m_isPrintingMapQualities;
- ostream* m_out;
- RefVector m_references;
- };
+// ---------------------------------------------
+// ConvertTool constants
+
+// supported conversion format command-line names
+static const string FORMAT_BED = "bed";
+static const string FORMAT_FASTA = "fasta";
+static const string FORMAT_FASTQ = "fastq";
+static const string FORMAT_JSON = "json";
+static const string FORMAT_SAM = "sam";
+static const string FORMAT_PILEUP = "pileup";
+static const string FORMAT_YAML = "yaml";
+
+// other constants
+static const unsigned int FASTA_LINE_MAX = 50;
+
+// ---------------------------------------------
+// ConvertPileupFormatVisitor declaration
+
+class ConvertPileupFormatVisitor : public PileupVisitor {
+
+ // ctor & dtor
+ public:
+ ConvertPileupFormatVisitor(const RefVector& references,
+ const string& fastaFilename,
+ const bool isPrintingMapQualities,
+ ostream* out);
+ ~ConvertPileupFormatVisitor(void);
+
+ // PileupVisitor interface implementation
+ public:
+ void Visit(const PileupPosition& pileupData);
+
+ // data members
+ private:
+ Fasta m_fasta;
+ bool m_hasFasta;
+ bool m_isPrintingMapQualities;
+ ostream* m_out;
+ RefVector m_references;
+};
} // namespace BamTools
struct ConvertTool::ConvertSettings {
- // flags
+ // flag
bool HasInput;
bool HasOutput;
bool HasFormat;
// ctor & dtor
public:
- ConvertToolPrivate(ConvertTool::ConvertSettings* settings);
- ~ConvertToolPrivate(void);
+ ConvertToolPrivate(ConvertTool::ConvertSettings* settings)
+ : m_settings(settings)
+ , m_out(cout.rdbuf())
+ { }
+
+ ~ConvertToolPrivate(void) { }
// interface
public:
// internal methods
private:
void PrintBed(const BamAlignment& a);
- void PrintBedGraph(const BamAlignment& a);
void PrintFasta(const BamAlignment& a);
void PrintFastq(const BamAlignment& a);
void PrintJson(const BamAlignment& a);
void PrintSam(const BamAlignment& a);
- void PrintWiggle(const BamAlignment& a);
+ void PrintYaml(const BamAlignment& a);
// special case - uses the PileupEngine
bool RunPileupConversion(BamMultiReader* reader);
ostream m_out;
};
-ConvertTool::ConvertToolPrivate::ConvertToolPrivate(ConvertTool::ConvertSettings* settings)
- : m_settings(settings)
- , m_out(cout.rdbuf()) // default output to cout
-{ }
-
-ConvertTool::ConvertToolPrivate::~ConvertToolPrivate(void) { }
-
bool ConvertTool::ConvertToolPrivate::Run(void) {
// ------------------------------------
// open input files
BamMultiReader reader;
- reader.Open(m_settings->InputFiles);
+ if ( !reader.Open(m_settings->InputFiles) ) {
+ cerr << "bamtools convert ERROR: could not open input BAM file(s)... Aborting." << endl;
+ return false;
+ }
+
+ // if input is not stdin & a region is provided, look for index files
+ if ( m_settings->HasInput && m_settings->HasRegion ) {
+ if ( !reader.LocateIndexes() ) {
+ cerr << "bamtools convert ERROR: could not locate index file(s)... Aborting." << endl;
+ return false;
+ }
+ }
+
+ // retrieve reference data
m_references = reader.GetReferenceData();
// set region if specified
BamRegion region;
if ( m_settings->HasRegion ) {
if ( Utilities::ParseRegionString(m_settings->Region, reader, region) ) {
- if ( !reader.SetRegion(region) ) {
- cerr << "Could not set BamReader region to REGION: " << m_settings->Region << endl;
- return false;
+
+ if ( reader.HasIndexes() ) {
+ if ( !reader.SetRegion(region) ) {
+ cerr << "bamtools convert ERROR: set region failed. Check that REGION describes a valid range" << endl;
+ reader.Close();
+ return false;
+ }
}
+
} else {
- cerr << "Could not parse REGION: " << m_settings->Region << endl;
+ cerr << "bamtools convert ERROR: could not parse REGION: " << m_settings->Region << endl;
+ cerr << "Check that REGION is in valid format (see documentation) and that the coordinates are valid"
+ << endl;
+ reader.Close();
return false;
}
}
// open output file stream
outFile.open(m_settings->OutputFilename.c_str());
if ( !outFile ) {
- cerr << "Could not open " << m_settings->OutputFilename << " for output" << endl;
+ cerr << "bamtools convert ERROR: could not open " << m_settings->OutputFilename
+ << " for output" << endl;
return false;
}
// set function pointer to proper conversion method
void (BamTools::ConvertTool::ConvertToolPrivate::*pFunction)(const BamAlignment&) = 0;
- if ( m_settings->Format == FORMAT_BED ) pFunction = &BamTools::ConvertTool::ConvertToolPrivate::PrintBed;
- else if ( m_settings->Format == FORMAT_BEDGRAPH ) pFunction = &BamTools::ConvertTool::ConvertToolPrivate::PrintBedGraph;
- else if ( m_settings->Format == FORMAT_FASTA ) pFunction = &BamTools::ConvertTool::ConvertToolPrivate::PrintFasta;
- else if ( m_settings->Format == FORMAT_FASTQ ) pFunction = &BamTools::ConvertTool::ConvertToolPrivate::PrintFastq;
- else if ( m_settings->Format == FORMAT_JSON ) pFunction = &BamTools::ConvertTool::ConvertToolPrivate::PrintJson;
- else if ( m_settings->Format == FORMAT_SAM ) pFunction = &BamTools::ConvertTool::ConvertToolPrivate::PrintSam;
- else if ( m_settings->Format == FORMAT_WIGGLE ) pFunction = &BamTools::ConvertTool::ConvertToolPrivate::PrintWiggle;
+ if ( m_settings->Format == FORMAT_BED ) pFunction = &BamTools::ConvertTool::ConvertToolPrivate::PrintBed;
+ else if ( m_settings->Format == FORMAT_FASTA ) pFunction = &BamTools::ConvertTool::ConvertToolPrivate::PrintFasta;
+ else if ( m_settings->Format == FORMAT_FASTQ ) pFunction = &BamTools::ConvertTool::ConvertToolPrivate::PrintFastq;
+ else if ( m_settings->Format == FORMAT_JSON ) pFunction = &BamTools::ConvertTool::ConvertToolPrivate::PrintJson;
+ else if ( m_settings->Format == FORMAT_SAM ) pFunction = &BamTools::ConvertTool::ConvertToolPrivate::PrintSam;
+ else if ( m_settings->Format == FORMAT_YAML ) pFunction = &BamTools::ConvertTool::ConvertToolPrivate::PrintYaml;
else {
- cerr << "Unrecognized format: " << m_settings->Format << endl;
- cerr << "Please see help|README (?) for details on supported formats " << endl;
+ cerr << "bamtools convert ERROR: unrecognized format: " << m_settings->Format << endl;
+ cerr << "Please see documentation for list of supported formats " << endl;
formatError = true;
convertedOk = false;
}
// iterate through file, doing conversion
BamAlignment a;
- while ( reader.GetNextAlignment(a) )
- (this->*pFunction)(a);
+ while ( reader.GetNextAlignment(a) )
+ (this->*pFunction)(a);
// set flag for successful conversion
convertedOk = true;
// ------------------------
// clean up & exit
-
reader.Close();
- if ( m_settings->HasOutput ) outFile.close();
+ if ( m_settings->HasOutput )
+ outFile.close();
return convertedOk;
}
m_out << m_references.at(a.RefID).RefName << "\t"
<< a.Position << "\t"
- << a.GetEndPosition() + 1 << "\t"
+ << a.GetEndPosition() << "\t"
<< a.Name << "\t"
<< a.MapQuality << "\t"
<< (a.IsReverseStrand() ? "-" : "+") << endl;
}
-void ConvertTool::ConvertToolPrivate::PrintBedGraph(const BamAlignment& a) {
- ;
-}
-
// print BamAlignment in FASTA format
// N.B. - uses QueryBases NOT AlignedBases
void ConvertTool::ConvertToolPrivate::PrintFasta(const BamAlignment& a) {
// >BamAlignment.Name
// BamAlignment.QueryBases (up to FASTA_LINE_MAX bases per line)
// ...
+ //
+ // N.B. - QueryBases are reverse-complemented if aligned to reverse strand
// print header
- m_out << "> " << a.Name << endl;
+ m_out << ">" << a.Name << endl;
+
+ // handle reverse strand alignment - bases
+ string sequence = a.QueryBases;
+ if ( a.IsReverseStrand() )
+ Utilities::ReverseComplement(sequence);
// if sequence fits on single line
- if ( a.QueryBases.length() <= FASTA_LINE_MAX )
- m_out << a.QueryBases << endl;
+ if ( sequence.length() <= FASTA_LINE_MAX )
+ m_out << sequence << endl;
// else split over multiple lines
else {
size_t position = 0;
- size_t seqLength = a.QueryBases.length();
+ size_t seqLength = sequence.length(); // handle reverse strand alignment - bases & qualitiesth();
// write subsequences to each line
while ( position < (seqLength - FASTA_LINE_MAX) ) {
- m_out << a.QueryBases.substr(position, FASTA_LINE_MAX) << endl;
+ m_out << sequence.substr(position, FASTA_LINE_MAX) << endl;
position += FASTA_LINE_MAX;
}
// write final subsequence
- m_out << a.QueryBases.substr(position) << endl;
+ m_out << sequence.substr(position) << endl;
}
}
// BamAlignment.QueryBases
// +
// BamAlignment.Qualities
+ //
+ // N.B. - QueryBases are reverse-complemented (& Qualities reversed) if aligned to reverse strand .
+ // Name is appended "/1" or "/2" if paired-end, to reflect which mate this entry is.
+
+ // handle paired-end alignments
+ string name = a.Name;
+ if ( a.IsPaired() )
+ name.append( (a.IsFirstMate() ? "/1" : "/2") );
+
+ // handle reverse strand alignment - bases & qualities
+ string qualities = a.Qualities;
+ string sequence = a.QueryBases;
+ if ( a.IsReverseStrand() ) {
+ Utilities::Reverse(qualities);
+ Utilities::ReverseComplement(sequence);
+ }
- m_out << "@" << a.Name << endl
- << a.QueryBases << endl
- << "+" << endl
- << a.Qualities << endl;
+ // write to output stream
+ m_out << "@" << name << endl
+ << sequence << endl
+ << "+" << endl
+ << qualities << endl;
}
// print BamAlignment in JSON format
if ( !cigarData.empty() ) {
m_out << "\"cigar\":[";
vector<CigarOp>::const_iterator cigarBegin = cigarData.begin();
- vector<CigarOp>::const_iterator cigarIter = cigarBegin;
- vector<CigarOp>::const_iterator cigarEnd = cigarData.end();
+ vector<CigarOp>::const_iterator cigarIter = cigarBegin;
+ vector<CigarOp>::const_iterator cigarEnd = cigarData.end();
for ( ; cigarIter != cigarEnd; ++cigarIter ) {
const CigarOp& op = (*cigarIter);
- if (cigarIter != cigarBegin) m_out << ",";
+ if (cigarIter != cigarBegin)
+ m_out << ",";
m_out << "\"" << op.Length << op.Type << "\"";
}
m_out << "],";
m_out << "\"queryBases\":\"" << a.QueryBases << "\",";
// write qualities
- if ( !a.Qualities.empty() ) {
+ if ( !a.Qualities.empty() && a.Qualities.at(0) != (char)0xFF ) {
string::const_iterator s = a.Qualities.begin();
m_out << "\"qualities\":[" << static_cast<short>(*s) - 33;
++s;
- for (; s != a.Qualities.end(); ++s) {
+ for ( ; s != a.Qualities.end(); ++s )
m_out << "," << static_cast<short>(*s) - 33;
- }
m_out << "],";
}
+ // write alignment's source BAM file
+ m_out << "\"filename\":" << a.Filename << ",";
+
// write tag data
const char* tagData = a.TagData.c_str();
const size_t tagDataLength = a.TagData.length();
size_t index = 0;
- if (index < tagDataLength) {
+ if ( index < tagDataLength ) {
m_out << "\"tags\":{";
while ( index < tagDataLength ) {
- if (index > 0)
+ if ( index > 0 )
m_out << ",";
// write tag name
// get data type
char type = a.TagData.at(index);
++index;
-
- switch (type) {
- case('A') :
+ switch ( type ) {
+ case (Constants::BAM_TAG_TYPE_ASCII) :
m_out << "\"" << tagData[index] << "\"";
++index;
break;
- case('C') :
- m_out << (int)tagData[index];
- ++index;
+ case (Constants::BAM_TAG_TYPE_INT8) :
+ // force value into integer-type (instead of char value)
+ m_out << static_cast<int16_t>(tagData[index]);
+ ++index;
break;
-
- case('c') :
- m_out << (int)tagData[index];
+
+ case (Constants::BAM_TAG_TYPE_UINT8) :
+ // force value into integer-type (instead of char value)
+ m_out << static_cast<uint16_t>(tagData[index]);
++index;
break;
- case('S') :
- m_out << BgzfData::UnpackUnsignedShort(&tagData[index]);
- index += 2;
- break;
-
- case('s') :
- m_out << BgzfData::UnpackSignedShort(&tagData[index]);
- index += 2;
+ case (Constants::BAM_TAG_TYPE_INT16) :
+ m_out << BamTools::UnpackSignedShort(&tagData[index]);
+ index += sizeof(int16_t);
break;
-
- case('I') :
- m_out << BgzfData::UnpackUnsignedInt(&tagData[index]);
- index += 4;
+
+ case (Constants::BAM_TAG_TYPE_UINT16) :
+ m_out << BamTools::UnpackUnsignedShort(&tagData[index]);
+ index += sizeof(uint16_t);
break;
-
- case('i') :
- m_out << BgzfData::UnpackSignedInt(&tagData[index]);
- index += 4;
+
+ case (Constants::BAM_TAG_TYPE_INT32) :
+ m_out << BamTools::UnpackSignedInt(&tagData[index]);
+ index += sizeof(int32_t);
break;
-
- case('f') :
- m_out << BgzfData::UnpackFloat(&tagData[index]);
- index += 4;
+
+ case (Constants::BAM_TAG_TYPE_UINT32) :
+ m_out << BamTools::UnpackUnsignedInt(&tagData[index]);
+ index += sizeof(uint32_t);
break;
-
- case('d') :
- m_out << BgzfData::UnpackDouble(&tagData[index]);
- index += 8;
+
+ case (Constants::BAM_TAG_TYPE_FLOAT) :
+ m_out << BamTools::UnpackFloat(&tagData[index]);
+ index += sizeof(float);
break;
- case('Z') :
- case('H') :
+ case (Constants::BAM_TAG_TYPE_HEX) :
+ case (Constants::BAM_TAG_TYPE_STRING) :
m_out << "\"";
while (tagData[index]) {
- m_out << tagData[index];
+ if (tagData[index] == '\"')
+ m_out << "\\\""; // escape for json
+ else
+ m_out << tagData[index];
++index;
}
m_out << "\"";
// <QNAME> <FLAG> <RNAME> <POS> <MAPQ> <CIGAR> <MRNM> <MPOS> <ISIZE> <SEQ> <QUAL> [ <TAG>:<VTYPE>:<VALUE> [...] ]
// write name & alignment flag
- m_out << a.Name << "\t" << a.AlignmentFlag << "\t";
-
+ m_out << a.Name << "\t" << a.AlignmentFlag << "\t";
+
// write reference name
if ( (a.RefID >= 0) && (a.RefID < (int)m_references.size()) )
m_out << m_references[a.RefID].RefName << "\t";
// write mate reference name, mate position, & insert size
if ( a.IsPaired() && (a.MateRefID >= 0) && (a.MateRefID < (int)m_references.size()) ) {
- if ( a.MateRefID == a.RefID ) m_out << "=\t";
- else m_out << m_references[a.MateRefID].RefName << "\t";
+ if ( a.MateRefID == a.RefID )
+ m_out << "=\t";
+ else
+ m_out << m_references[a.MateRefID].RefName << "\t";
m_out << a.MatePosition+1 << "\t" << a.InsertSize << "\t";
}
- else m_out << "*\t0\t0\t";
+ else
+ m_out << "*\t0\t0\t";
// write sequence
- if ( a.QueryBases.empty() ) m_out << "*\t";
- else m_out << a.QueryBases << "\t";
+ if ( a.QueryBases.empty() )
+ m_out << "*\t";
+ else
+ m_out << a.QueryBases << "\t";
// write qualities
- if ( a.Qualities.empty() ) m_out << "*";
- else m_out << a.Qualities;
+ if ( a.Qualities.empty() || (a.Qualities.at(0) == (char)0xFF) )
+ m_out << "*";
+ else
+ m_out << a.Qualities;
// write tag data
const char* tagData = a.TagData.c_str();
// get data type
char type = a.TagData.at(index);
++index;
- switch (type) {
- case('A') :
- m_out << "A:" << tagData[index];
- ++index;
- break;
-
- case('C') :
- m_out << "i:" << (int)tagData[index];
- ++index;
+ switch ( type ) {
+ case (Constants::BAM_TAG_TYPE_ASCII) :
+ m_out << "A:" << tagData[index];
+ ++index;
break;
-
- case('c') :
- m_out << "i:" << (int)tagData[index];
- ++index;
+
+ case (Constants::BAM_TAG_TYPE_INT8) :
+ // force value into integer-type (instead of char value)
+ m_out << "i:" << static_cast<int16_t>(tagData[index]);
+ ++index;
break;
-
- case('S') :
- m_out << "i:" << BgzfData::UnpackUnsignedShort(&tagData[index]);
- index += 2;
+
+ case (Constants::BAM_TAG_TYPE_UINT8) :
+ // force value into integer-type (instead of char value)
+ m_out << "i:" << static_cast<uint16_t>(tagData[index]);
+ ++index;
break;
-
- case('s') :
- m_out << "i:" << BgzfData::UnpackSignedShort(&tagData[index]);
- index += 2;
+
+ case (Constants::BAM_TAG_TYPE_INT16) :
+ m_out << "i:" << BamTools::UnpackSignedShort(&tagData[index]);
+ index += sizeof(int16_t);
break;
-
- case('I') :
- m_out << "i:" << BgzfData::UnpackUnsignedInt(&tagData[index]);
- index += 4;
+
+ case (Constants::BAM_TAG_TYPE_UINT16) :
+ m_out << "i:" << BamTools::UnpackUnsignedShort(&tagData[index]);
+ index += sizeof(uint16_t);
break;
-
- case('i') :
- m_out << "i:" << BgzfData::UnpackSignedInt(&tagData[index]);
- index += 4;
+
+ case (Constants::BAM_TAG_TYPE_INT32) :
+ m_out << "i:" << BamTools::UnpackSignedInt(&tagData[index]);
+ index += sizeof(int32_t);
break;
-
- case('f') :
- m_out << "f:" << BgzfData::UnpackFloat(&tagData[index]);
- index += 4;
+
+ case (Constants::BAM_TAG_TYPE_UINT32) :
+ m_out << "i:" << BamTools::UnpackUnsignedInt(&tagData[index]);
+ index += sizeof(uint32_t);
break;
-
- case('d') :
- m_out << "d:" << BgzfData::UnpackDouble(&tagData[index]);
- index += 8;
+
+ case (Constants::BAM_TAG_TYPE_FLOAT) :
+ m_out << "f:" << BamTools::UnpackFloat(&tagData[index]);
+ index += sizeof(float);
break;
-
- case('Z') :
- case('H') :
+
+ case (Constants::BAM_TAG_TYPE_HEX) : // fall-through
+ case (Constants::BAM_TAG_TYPE_STRING) :
m_out << type << ":";
while (tagData[index]) {
m_out << tagData[index];
++index;
}
- ++index;
- break;
+ ++index;
+ break;
}
- if ( tagData[index] == '\0')
+ if ( tagData[index] == '\0' )
break;
}
m_out << endl;
}
-void ConvertTool::ConvertToolPrivate::PrintWiggle(const BamAlignment& a) {
- ;
+// Print BamAlignment in YAML format
+void ConvertTool::ConvertToolPrivate::PrintYaml(const BamAlignment& a) {
+
+ // write alignment name
+ m_out << "---" << endl;
+ m_out << a.Name << ":" << endl;
+
+ // write alignment data
+ m_out << " " << "AlndBases: " << a.AlignedBases << endl;
+ m_out << " " << "Qualities: " << a.Qualities << endl;
+ m_out << " " << "Name: " << a.Name << endl;
+ m_out << " " << "Length: " << a.Length << endl;
+ m_out << " " << "TagData: " << a.TagData << endl;
+ m_out << " " << "RefID: " << a.RefID << endl;
+ m_out << " " << "RefName: " << m_references[a.RefID].RefName << endl;
+ m_out << " " << "Position: " << a.Position << endl;
+ m_out << " " << "Bin: " << a.Bin << endl;
+ m_out << " " << "MapQuality: " << a.MapQuality << endl;
+ m_out << " " << "AlignmentFlag: " << a.AlignmentFlag << endl;
+ m_out << " " << "MateRefID: " << a.MateRefID << endl;
+ m_out << " " << "MatePosition: " << a.MatePosition << endl;
+ m_out << " " << "InsertSize: " << a.InsertSize << endl;
+ m_out << " " << "Filename: " << a.Filename << endl;
+
+ // write Cigar data
+ const vector<CigarOp>& cigarData = a.CigarData;
+ if ( !cigarData.empty() ) {
+ m_out << " " << "Cigar: ";
+ vector<CigarOp>::const_iterator cigarBegin = cigarData.begin();
+ vector<CigarOp>::const_iterator cigarIter = cigarBegin;
+ vector<CigarOp>::const_iterator cigarEnd = cigarData.end();
+ for ( ; cigarIter != cigarEnd; ++cigarIter ) {
+ const CigarOp& op = (*cigarIter);
+ m_out << op.Length << op.Type;
+ }
+ m_out << endl;
+ }
}
bool ConvertTool::ConvertToolPrivate::RunPileupConversion(BamMultiReader* reader) {
, m_impl(0)
{
// set program details
- Options::SetProgramInfo("bamtools convert", "converts BAM to a number of other formats", "-format <FORMAT> [-in <filename> -in <filename> ...] [-out <filename>] [other options]");
+ Options::SetProgramInfo("bamtools convert", "converts BAM to a number of other formats", "-format <FORMAT> [-in <filename> -in <filename> ...] [-out <filename>] [-region <REGION>] [format-specific options]");
// set up options
OptionGroup* IO_Opts = Options::CreateOptionGroup("Input & Output");
Options::AddValueOption("-in", "BAM filename", "the input BAM file(s)", "", m_settings->HasInput, m_settings->InputFiles, IO_Opts, Options::StandardIn());
Options::AddValueOption("-out", "BAM filename", "the output BAM file", "", m_settings->HasOutput, m_settings->OutputFilename, IO_Opts, Options::StandardOut());
Options::AddValueOption("-format", "FORMAT", "the output file format - see README for recognized formats", "", m_settings->HasFormat, m_settings->Format, IO_Opts);
-
- OptionGroup* FilterOpts = Options::CreateOptionGroup("Filters");
- Options::AddValueOption("-region", "REGION", "genomic region. Index file is recommended for better performance, and is read automatically if it exists as <filename>.bai. See \'bamtools help index\' for more details on creating one", "", m_settings->HasRegion, m_settings->Region, FilterOpts);
+ Options::AddValueOption("-region", "REGION", "genomic region. Index file is recommended for better performance, and is used automatically if it exists. See \'bamtools help index\' for more details on creating one", "", m_settings->HasRegion, m_settings->Region, IO_Opts);
OptionGroup* PileupOpts = Options::CreateOptionGroup("Pileup Options");
- Options::AddValueOption("-fasta", "FASTA filename", "FASTA reference file", "", m_settings->HasFastaFilename, m_settings->FastaFilename, PileupOpts, "");
+ Options::AddValueOption("-fasta", "FASTA filename", "FASTA reference file", "", m_settings->HasFastaFilename, m_settings->FastaFilename, PileupOpts);
Options::AddOption("-mapqual", "print the mapping qualities", m_settings->IsPrintingPileupMapQualities, PileupOpts);
OptionGroup* SamOpts = Options::CreateOptionGroup("SAM Options");
}
ConvertTool::~ConvertTool(void) {
+
delete m_settings;
m_settings = 0;
// parse command line arguments
Options::Parse(argc, argv, 1);
- // run internal ConvertTool implementation, return success/fail
+ // initialize ConvertTool with settings
m_impl = new ConvertToolPrivate(m_settings);
+ // run ConvertTool, return success/fail
if ( m_impl->Run() )
return 0;
else
char referenceBase('N');
if ( m_hasFasta && (pileupData.Position < m_references[pileupData.RefId].RefLength) ) {
if ( !m_fasta.GetBase(pileupData.RefId, pileupData.Position, referenceBase ) ) {
- cerr << "Pileup error : Could not read reference base from FASTA file" << endl;
+ cerr << "bamtools convert ERROR: pileup conversion - could not read reference base from FASTA file" << endl;
return;
}
}
toupper(base) == toupper(referenceBase) ||
tolower(base) == tolower(referenceBase) )
{
- base = (ba.IsReverseStrand() ? ',' : '.' );
+ base = ( ba.IsReverseStrand() ? ',' : '.' );
}
// mismatches reference
- else base = (ba.IsReverseStrand() ? tolower(base) : toupper(base) );
+ else base = ( ba.IsReverseStrand() ? tolower(base) : toupper(base) );
// store base
bases << base;
char deletedBase('N');
if ( m_hasFasta && (pileupData.Position+i < m_references[pileupData.RefId].RefLength) ) {
if ( !m_fasta.GetBase(pileupData.RefId, pileupData.Position+i, deletedBase ) ) {
- cerr << "Pileup error : Could not read reference base from FASTA file" << endl;
+ cerr << "bamtools convert ERROR: pileup conversion - could not read reference base from FASTA file" << endl;
return;
}
}
else bases << '*';
// if end of read segment
- if ( pa.IsSegmentEnd ) bases << '$';
+ if ( pa.IsSegmentEnd )
+ bases << '$';
// store current base quality
baseQualities << ba.Qualities.at(pa.PositionInAlignment);
projects / bamtools.git / blobdiff