// Marth Lab, Department of Biology, Boston College
// All rights reserved.
// ---------------------------------------------------------------------------
-// Last modified: 20 July 2010 (DB)
+// Last modified: 3 September 2010 (DB)
// ---------------------------------------------------------------------------
// Uses BGZF routines were adapted from the bgzf.c code developed at the Broad
// Institute.
// C++ includes
#include <algorithm>
+#include <fstream>
+#include <iostream>
#include <iterator>
+#include <sstream>
#include <string>
#include <vector>
-#include <iostream>
-#include <sstream>
// BamTools includes
#include "BGZF.h"
using namespace BamTools;
using namespace std;
+namespace BamTools {
+
+bool FileExists(const string& filename) {
+ ifstream f(filename.c_str(), ifstream::in);
+ return !f.fail();
+}
+
+} // namespace BamTools
+
// -----------------------------------------------------
// BamMultiReader implementation
// -----------------------------------------------------
// destructor
BamMultiReader::~BamMultiReader(void) {
- Close(); // close the bam files
- // clean up reader objects
- for (vector<pair<BamReader*, BamAlignment*> >::iterator it = readers.begin(); it != readers.end(); ++it) {
- delete it->first;
- delete it->second;
- }
+ Close();
}
// close the BAM files
void BamMultiReader::Close(void) {
+
+ // close all BAM readers and clean up pointers
+ vector<pair<BamReader*, BamAlignment*> >::iterator readerIter = readers.begin();
+ vector<pair<BamReader*, BamAlignment*> >::iterator readerEnd = readers.end();
+ for ( ; readerIter != readerEnd; ++readerIter) {
+
+ BamReader* reader = (*readerIter).first;
+ BamAlignment* alignment = (*readerIter).second;
+
+ // close the reader
+ if ( reader) reader->Close();
+
+ // delete reader pointer
+ delete reader;
+ reader = 0;
+
+ // delete alignment pointer
+ delete alignment;
+ alignment = 0;
+ }
+
+ // clear out the container
+ readers.clear();
+}
+
+// saves index data to BAM index files (".bai"/".bti") where necessary, returns success/fail
+bool BamMultiReader::CreateIndexes(bool useDefaultIndex) {
+ bool result = true;
for (vector<pair<BamReader*, BamAlignment*> >::iterator it = readers.begin(); it != readers.end(); ++it) {
BamReader* reader = it->first;
- reader->Close(); // close the reader
+ result &= reader->CreateIndex(useDefaultIndex);
}
+ return result;
}
-// updates the reference id stored in the BamMultiReader
-// to reflect the current state of the readers
-void BamMultiReader::UpdateReferenceID(void) {
- // the alignments are sorted by position, so the first alignment will always have the lowest reference ID
- if (alignments.begin()->second.second->RefID != CurrentRefID) {
- // get the next reference id
- // while there aren't any readers at the next ref id
- // increment the ref id
- int nextRefID = CurrentRefID;
- while (alignments.begin()->second.second->RefID != nextRefID) {
- ++nextRefID;
- }
- //cerr << "updating reference id from " << CurrentRefID << " to " << nextRefID << endl;
- CurrentRefID = nextRefID;
+// for debugging
+void BamMultiReader::DumpAlignmentIndex(void) {
+ for (AlignmentIndex::const_iterator it = alignments.begin(); it != alignments.end(); ++it) {
+ cerr << it->first.first << ":" << it->first.second << " " << it->second.first->GetFilename() << endl;
}
}
-// checks if any readers still have alignments
-bool BamMultiReader::HasOpenReaders() {
- return alignments.size() > 0;
+// makes a virtual, unified header for all the bam files in the multireader
+const string BamMultiReader::GetHeaderText(void) const {
+
+ string mergedHeader = "";
+ map<string, bool> readGroups;
+
+ // foreach extraction entry (each BAM file)
+ for (vector<pair<BamReader*, BamAlignment*> >::const_iterator rs = readers.begin(); rs != readers.end(); ++rs) {
+
+ map<string, bool> currentFileReadGroups;
+
+ BamReader* reader = rs->first;
+
+ stringstream header(reader->GetHeaderText());
+ vector<string> lines;
+ string item;
+ while (getline(header, item))
+ lines.push_back(item);
+
+ for (vector<string>::const_iterator it = lines.begin(); it != lines.end(); ++it) {
+
+ // get next line from header, skip if empty
+ string headerLine = *it;
+ if ( headerLine.empty() ) { continue; }
+
+ // if first file, save HD & SQ entries
+ if ( rs == readers.begin() ) {
+ if ( headerLine.find("@HD") == 0 || headerLine.find("@SQ") == 0) {
+ mergedHeader.append(headerLine.c_str());
+ mergedHeader.append(1, '\n');
+ }
+ }
+
+ // (for all files) append RG entries if they are unique
+ if ( headerLine.find("@RG") == 0 ) {
+ stringstream headerLineSs(headerLine);
+ string part, readGroupPart, readGroup;
+ while(std::getline(headerLineSs, part, '\t')) {
+ stringstream partSs(part);
+ string subtag;
+ std::getline(partSs, subtag, ':');
+ if (subtag == "ID") {
+ std::getline(partSs, readGroup, ':');
+ break;
+ }
+ }
+ if (readGroups.find(readGroup) == readGroups.end()) { // prevents duplicate @RG entries
+ mergedHeader.append(headerLine.c_str() );
+ mergedHeader.append(1, '\n');
+ readGroups[readGroup] = true;
+ currentFileReadGroups[readGroup] = true;
+ } else {
+ // warn iff we are reading one file and discover duplicated @RG tags in the header
+ // otherwise, we emit no warning, as we might be merging multiple BAM files with identical @RG tags
+ if (currentFileReadGroups.find(readGroup) != currentFileReadGroups.end()) {
+ cerr << "WARNING: duplicate @RG tag " << readGroup
+ << " entry in header of " << reader->GetFilename() << endl;
+ }
+ }
+ }
+ }
+ }
+
+ // return merged header text
+ return mergedHeader;
}
// get next alignment among all files
}
+// ---------------------------------------------------------------------------------------
+//
+// NB: The following GetReferenceX() functions assume that we have identical
+// references for all BAM files. We enforce this by invoking the above
+// validation function (ValidateReaders) to verify that our reference data
+// is the same across all files on Open, so we will not encounter a situation
+// in which there is a mismatch and we are still live.
+//
+// ---------------------------------------------------------------------------------------
+
+// returns the number of reference sequences
+const int BamMultiReader::GetReferenceCount(void) const {
+ return readers.front().first->GetReferenceCount();
+}
+
+// returns vector of reference objects
+const BamTools::RefVector BamMultiReader::GetReferenceData(void) const {
+ return readers.front().first->GetReferenceData();
+}
+
+// returns refID from reference name
+const int BamMultiReader::GetReferenceID(const string& refName) const {
+ return readers.front().first->GetReferenceID(refName);
+}
+
+// ---------------------------------------------------------------------------------------
+
+// checks if any readers still have alignments
+bool BamMultiReader::HasOpenReaders() {
+ return alignments.size() > 0;
+}
+
+// returns whether underlying BAM readers ALL have an index loaded
+// this is useful to indicate whether Jump() or SetRegion() are possible
+bool BamMultiReader::IsIndexLoaded(void) const {
+ bool ok = true;
+ vector<pair<BamReader*, BamAlignment*> >::const_iterator readerIter = readers.begin();
+ vector<pair<BamReader*, BamAlignment*> >::const_iterator readerEnd = readers.end();
+ for ( ; readerIter != readerEnd; ++readerIter ) {
+ const BamReader* reader = (*readerIter).first;
+ if ( reader ) ok &= reader->IsIndexLoaded();
+ }
+ return ok;
+}
+
// jumps to specified region(refID, leftBound) in BAM files, returns success/fail
bool BamMultiReader::Jump(int refID, int position) {
return result;
}
-bool BamMultiReader::SetRegion(const int& leftRefID, const int& leftPosition, const int& rightRefID, const int& rightPosition) {
-
- BamRegion region(leftRefID, leftPosition, rightRefID, rightPosition);
-
- return SetRegion(region);
-
-}
-
-bool BamMultiReader::SetRegion(const BamRegion& region) {
-
- Region = region;
-
- // NB: While it may make sense to track readers in which we can
- // successfully SetRegion, In practice a failure of SetRegion means "no
- // alignments here." It makes sense to simply accept the failure,
- // UpdateAlignments(), and continue.
-
- for (vector<pair<BamReader*, BamAlignment*> >::iterator it = readers.begin(); it != readers.end(); ++it) {
- it->first->SetRegion(region);
- }
-
- UpdateAlignments();
-
- return true;
-
-}
-
-void BamMultiReader::UpdateAlignments(void) {
- // Update Alignments
- alignments.clear();
- for (vector<pair<BamReader*, BamAlignment*> >::iterator it = readers.begin(); it != readers.end(); ++it) {
- BamReader* br = it->first;
- BamAlignment* ba = it->second;
- if (br->GetNextAlignment(*ba)) {
- alignments.insert(make_pair(make_pair(ba->RefID, ba->Position),
- make_pair(br, ba)));
- } else {
- // assume BamReader end of region / EOF
- }
- }
-}
-
// opens BAM files
bool BamMultiReader::Open(const vector<string> filenames, bool openIndexes, bool coreMode, bool useDefaultIndex) {
// for filename in filenames
fileNames = filenames; // save filenames in our multireader
for (vector<string>::const_iterator it = filenames.begin(); it != filenames.end(); ++it) {
- string filename = *it;
+
+ const string filename = *it;
BamReader* reader = new BamReader;
bool openedOK = true;
if (openIndexes) {
- if (useDefaultIndex)
- openedOK = reader->Open(filename, filename + ".bai");
- else
- openedOK = reader->Open(filename, filename + ".bti");
- } else {
- openedOK = reader->Open(filename); // for merging, jumping is disallowed
- }
+
+ const string defaultIndexFilename = filename + ".bai";
+ const string bamToolsIndexFilename = filename + ".bti";
+
+ // if default BAM index requested and one exists
+ if ( useDefaultIndex && FileExists(defaultIndexFilename) )
+ openedOK = reader->Open(filename, defaultIndexFilename);
+
+ // else see if BamTools index exists
+ else if ( FileExists(bamToolsIndexFilename) )
+ openedOK = reader->Open(filename, bamToolsIndexFilename);
+
+ // else none exist... just open without
+ else
+ openedOK = reader->Open(filename);
+ }
+
+ // ignoring index file(s)
+ else openedOK = reader->Open(filename);
// if file opened ok, check that it can be read
if ( openedOK ) {
bool fileOK = true;
BamAlignment* alignment = new BamAlignment;
- if (coreMode) {
- fileOK &= reader->GetNextAlignmentCore(*alignment);
- } else {
- fileOK &= reader->GetNextAlignment(*alignment);
- }
+ fileOK &= ( coreMode ? reader->GetNextAlignmentCore(*alignment) : reader->GetNextAlignment(*alignment) );
if (fileOK) {
readers.push_back(make_pair(reader, alignment)); // store pointers to our readers for cleanup
// if only file available & could not be read, return failure
if ( filenames.size() == 1 ) return false;
}
-
}
- // TODO; any more error handling on openedOK ??
+ // TODO; any more error handling when openedOKvis false ??
else
return false;
}
}
}
-// for debugging
-void BamMultiReader::DumpAlignmentIndex(void) {
- for (AlignmentIndex::const_iterator it = alignments.begin(); it != alignments.end(); ++it) {
- cerr << it->first.first << ":" << it->first.second << " " << it->second.first->GetFilename() << endl;
- }
-}
-
// returns BAM file pointers to beginning of alignment data
bool BamMultiReader::Rewind(void) {
bool result = true;
return result;
}
-// saves index data to BAM index files (".bai"/".bti") where necessary, returns success/fail
-bool BamMultiReader::CreateIndexes(bool useDefaultIndex) {
- bool result = true;
- for (vector<pair<BamReader*, BamAlignment*> >::iterator it = readers.begin(); it != readers.end(); ++it) {
- BamReader* reader = it->first;
- result &= reader->CreateIndex(useDefaultIndex);
- }
- return result;
+bool BamMultiReader::SetRegion(const int& leftRefID, const int& leftPosition, const int& rightRefID, const int& rightPosition) {
+ BamRegion region(leftRefID, leftPosition, rightRefID, rightPosition);
+ return SetRegion(region);
}
-// makes a virtual, unified header for all the bam files in the multireader
-const string BamMultiReader::GetHeaderText(void) const {
-
- string mergedHeader = "";
- map<string, bool> readGroups;
-
- // foreach extraction entry (each BAM file)
- for (vector<pair<BamReader*, BamAlignment*> >::const_iterator rs = readers.begin(); rs != readers.end(); ++rs) {
-
- map<string, bool> currentFileReadGroups;
-
- BamReader* reader = rs->first;
+bool BamMultiReader::SetRegion(const BamRegion& region) {
- stringstream header(reader->GetHeaderText());
- vector<string> lines;
- string item;
- while (getline(header, item))
- lines.push_back(item);
+ Region = region;
- for (vector<string>::const_iterator it = lines.begin(); it != lines.end(); ++it) {
+ // NB: While it may make sense to track readers in which we can
+ // successfully SetRegion, In practice a failure of SetRegion means "no
+ // alignments here." It makes sense to simply accept the failure,
+ // UpdateAlignments(), and continue.
- // get next line from header, skip if empty
- string headerLine = *it;
- if ( headerLine.empty() ) { continue; }
+ for (vector<pair<BamReader*, BamAlignment*> >::iterator it = readers.begin(); it != readers.end(); ++it) {
+ it->first->SetRegion(region);
+ }
- // if first file, save HD & SQ entries
- if ( rs == readers.begin() ) {
- if ( headerLine.find("@HD") == 0 || headerLine.find("@SQ") == 0) {
- mergedHeader.append(headerLine.c_str());
- mergedHeader.append(1, '\n');
- }
- }
+ UpdateAlignments();
+ return true;
+}
- // (for all files) append RG entries if they are unique
- if ( headerLine.find("@RG") == 0 ) {
- stringstream headerLineSs(headerLine);
- string part, readGroupPart, readGroup;
- while(std::getline(headerLineSs, part, '\t')) {
- stringstream partSs(part);
- string subtag;
- std::getline(partSs, subtag, ':');
- if (subtag == "ID") {
- std::getline(partSs, readGroup, ':');
- break;
- }
- }
- if (readGroups.find(readGroup) == readGroups.end()) { // prevents duplicate @RG entries
- mergedHeader.append(headerLine.c_str() );
- mergedHeader.append(1, '\n');
- readGroups[readGroup] = true;
- currentFileReadGroups[readGroup] = true;
- } else {
- // warn iff we are reading one file and discover duplicated @RG tags in the header
- // otherwise, we emit no warning, as we might be merging multiple BAM files with identical @RG tags
- if (currentFileReadGroups.find(readGroup) != currentFileReadGroups.end()) {
- cerr << "WARNING: duplicate @RG tag " << readGroup
- << " entry in header of " << reader->GetFilename() << endl;
- }
- }
- }
+void BamMultiReader::UpdateAlignments(void) {
+ // Update Alignments
+ alignments.clear();
+ for (vector<pair<BamReader*, BamAlignment*> >::iterator it = readers.begin(); it != readers.end(); ++it) {
+ BamReader* br = it->first;
+ BamAlignment* ba = it->second;
+ if (br->GetNextAlignment(*ba)) {
+ alignments.insert(make_pair(make_pair(ba->RefID, ba->Position),
+ make_pair(br, ba)));
+ } else {
+ // assume BamReader end of region / EOF
}
}
+}
- // return merged header text
- return mergedHeader;
+// updates the reference id stored in the BamMultiReader
+// to reflect the current state of the readers
+void BamMultiReader::UpdateReferenceID(void) {
+ // the alignments are sorted by position, so the first alignment will always have the lowest reference ID
+ if (alignments.begin()->second.second->RefID != CurrentRefID) {
+ // get the next reference id
+ // while there aren't any readers at the next ref id
+ // increment the ref id
+ int nextRefID = CurrentRefID;
+ while (alignments.begin()->second.second->RefID != nextRefID) {
+ ++nextRefID;
+ }
+ //cerr << "updating reference id from " << CurrentRefID << " to " << nextRefID << endl;
+ CurrentRefID = nextRefID;
+ }
}
// ValidateReaders checks that all the readers point to BAM files representing
}
}
}
-
-// NB: The following functions assume that we have identical references for all
-// BAM files. We enforce this by invoking the above validation function
-// (ValidateReaders) to verify that our reference data is the same across all
-// files on Open, so we will not encounter a situation in which there is a
-// mismatch and we are still live.
-
-// returns the number of reference sequences
-const int BamMultiReader::GetReferenceCount(void) const {
- return readers.front().first->GetReferenceCount();
-}
-
-// returns vector of reference objects
-const BamTools::RefVector BamMultiReader::GetReferenceData(void) const {
- return readers.front().first->GetReferenceData();
-}
-
-const int BamMultiReader::GetReferenceID(const string& refName) const {
- return readers.front().first->GetReferenceID(refName);
-}