1 // ***************************************************************************
2 // BamMultiReader_p.cpp (c) 2010 Derek Barnett, Erik Garrison
3 // Marth Lab, Department of Biology, Boston College
4 // All rights reserved.
5 // ---------------------------------------------------------------------------
6 // Last modified: 23 December 2010 (DB)
7 // ---------------------------------------------------------------------------
8 // Functionality for simultaneously reading multiple BAM files
9 // *************************************************************************
11 #include <api/BamAlignment.h>
12 #include <api/BamMultiReader.h>
13 #include <api/internal/BamMultiMerger_p.h>
14 #include <api/internal/BamMultiReader_p.h>
15 using namespace BamTools;
16 using namespace BamTools::Internal;
26 BamMultiReaderPrivate::BamMultiReaderPrivate(void)
29 , m_isSortedByPosition(true)
33 BamMultiReaderPrivate::~BamMultiReaderPrivate(void) {
35 // close all open BAM readers
38 // clean up alignment cache
43 // close the BAM files
44 void BamMultiReaderPrivate::Close(void) {
46 // clear out alignment cache
47 m_alignments->Clear();
49 // iterate over readers
50 vector<ReaderAlignment>::iterator readerIter = m_readers.begin();
51 vector<ReaderAlignment>::iterator readerEnd = m_readers.end();
52 for ( ; readerIter != readerEnd; ++readerIter ) {
55 BamReader* reader = (*readerIter).first;
56 BamAlignment* alignment = (*readerIter).second;
57 if ( reader ) reader->Close();
71 m_isSortedByPosition = true;
74 // saves index data to BAM index files (".bai"/".bti") where necessary, returns success/fail
75 bool BamMultiReaderPrivate::CreateIndexes(bool useStandardIndex) {
78 vector<ReaderAlignment>::iterator readerIter = m_readers.begin();
79 vector<ReaderAlignment>::iterator readerEnd = m_readers.end();
80 for ( ; readerIter != readerEnd; ++readerIter ) {
81 BamReader* reader = (*readerIter).first;
82 result &= reader->CreateIndex(useStandardIndex);
87 const string BamMultiReaderPrivate::ExtractReadGroup(const string& headerLine) const {
90 stringstream headerLineSs(headerLine);
93 // parse @RG header line, looking for the ID: tag
94 while( getline(headerLineSs, part, '\t') ) {
95 stringstream partSs(part);
97 getline(partSs, subtag, ':');
98 if ( subtag == "ID" ) {
99 getline(partSs, readGroup, ':');
106 // makes a virtual, unified header for all the bam files in the multireader
107 const string BamMultiReaderPrivate::GetHeaderText(void) const {
109 // just spit single header out if only have one reader open
110 if ( m_readers.size() == 1 ) {
113 vector<ReaderAlignment>::const_iterator readerBegin = m_readers.begin();
114 const ReaderAlignment& entry = (*readerBegin);
115 const BamReader* reader = entry.first;
116 if ( reader == 0 ) return "";
117 return reader->GetHeaderText();
120 string mergedHeader = "";
121 map<string, bool> readGroups;
123 // foreach extraction entry (each BAM file)
124 vector<ReaderAlignment>::const_iterator readerBegin = m_readers.begin();
125 vector<ReaderAlignment>::const_iterator readerIter = readerBegin;
126 vector<ReaderAlignment>::const_iterator readerEnd = m_readers.end();
127 for ( ; readerIter != readerEnd; ++readerIter ) {
129 // get header from reader
130 const BamReader* reader = (*readerIter).first;
131 if ( reader == 0 ) continue;
132 string headerText = reader->GetHeaderText();
133 if ( headerText.empty() ) continue;
135 // store header text in lines
136 map<string, bool> currentFileReadGroups;
137 const vector<string> lines = SplitHeaderText(headerText);
139 // iterate over header lines
140 vector<string>::const_iterator linesIter = lines.begin();
141 vector<string>::const_iterator linesEnd = lines.end();
142 for ( ; linesIter != linesEnd; ++linesIter ) {
144 // get next line from header, skip if empty
145 const string headerLine = (*linesIter);
146 if ( headerLine.empty() ) { continue; }
148 // if first file, save HD & SQ entries
149 if ( readerIter == readerBegin ) {
150 if ( headerLine.find("@HD") == 0 || headerLine.find("@SQ") == 0) {
151 mergedHeader.append(headerLine.c_str());
152 mergedHeader.append(1, '\n');
156 // (for all files) append RG entries if they are unique
157 if ( headerLine.find("@RG") == 0 ) {
159 // extract read group name from line
160 const string readGroup = ExtractReadGroup(headerLine);
162 // make sure not to duplicate @RG entries
163 if ( readGroups.find(readGroup) == readGroups.end() ) {
164 mergedHeader.append(headerLine.c_str() );
165 mergedHeader.append(1, '\n');
166 readGroups[readGroup] = true;
167 currentFileReadGroups[readGroup] = true;
169 // warn iff we are reading one file and discover duplicated @RG tags in the header
170 // otherwise, we emit no warning, as we might be merging multiple BAM files with identical @RG tags
171 if ( currentFileReadGroups.find(readGroup) != currentFileReadGroups.end() ) {
172 cerr << "WARNING: duplicate @RG tag " << readGroup
173 << " entry in header of " << reader->GetFilename() << endl;
180 // return merged header text
184 // get next alignment among all files
185 bool BamMultiReaderPrivate::GetNextAlignment(BamAlignment& al) {
186 return LoadNextAlignment(al, false);
189 // get next alignment among all files without parsing character data from alignments
190 bool BamMultiReaderPrivate::GetNextAlignmentCore(BamAlignment& al) {
191 return LoadNextAlignment(al, true);
194 // ---------------------------------------------------------------------------------------
196 // NB: The following GetReferenceX() functions assume that we have identical
197 // references for all BAM files. We enforce this by invoking the above
198 // validation function (ValidateReaders) to verify that our reference data
199 // is the same across all files on Open, so we will not encounter a situation
200 // in which there is a mismatch and we are still live.
202 // ---------------------------------------------------------------------------------------
204 // returns the number of reference sequences
205 const int BamMultiReaderPrivate::GetReferenceCount(void) const {
206 const ReaderAlignment& firstReader = m_readers.front();
207 const BamReader* reader = firstReader.first;
208 if ( reader ) return reader->GetReferenceCount();
212 // returns vector of reference objects
213 const RefVector BamMultiReaderPrivate::GetReferenceData(void) const {
214 const ReaderAlignment& firstReader = m_readers.front();
215 const BamReader* reader = firstReader.first;
216 if ( reader ) return reader->GetReferenceData();
217 else return RefVector();
220 // returns refID from reference name
221 const int BamMultiReaderPrivate::GetReferenceID(const string& refName) const {
222 const ReaderAlignment& firstReader = m_readers.front();
223 const BamReader* reader = firstReader.first;
224 if ( reader ) return reader->GetReferenceID(refName);
225 else return -1; // ERROR case - how to report
228 // ---------------------------------------------------------------------------------------
230 // checks if any readers still have alignments
231 bool BamMultiReaderPrivate::HasOpenReaders(void) {
232 return ( m_alignments->Size() > 0 );
235 // returns whether underlying BAM readers ALL have an index loaded
236 // this is useful to indicate whether Jump() or SetRegion() are possible
237 bool BamMultiReaderPrivate::IsIndexLoaded(void) const {
239 vector<ReaderAlignment>::const_iterator readerIter = m_readers.begin();
240 vector<ReaderAlignment>::const_iterator readerEnd = m_readers.end();
241 for ( ; readerIter != readerEnd; ++readerIter ) {
242 const BamReader* reader = (*readerIter).first;
243 if ( reader ) ok &= reader->IsIndexLoaded();
248 // jumps to specified region(refID, leftBound) in BAM files, returns success/fail
249 bool BamMultiReaderPrivate::Jump(int refID, int position) {
252 vector<ReaderAlignment>::iterator readerIter = m_readers.begin();
253 vector<ReaderAlignment>::iterator readerEnd = m_readers.end();
254 for ( ; readerIter != readerEnd; ++readerIter ) {
255 BamReader* reader = (*readerIter).first;
256 if ( reader == 0 ) continue;
258 ok &= reader->Jump(refID, position);
260 cerr << "ERROR: could not jump " << reader->GetFilename()
261 << " to " << refID << ":" << position << endl;
266 if (ok) UpdateAlignments();
270 bool BamMultiReaderPrivate::LoadNextAlignment(BamAlignment& al, bool coreMode) {
272 // bail out if no more data to process
273 if ( !HasOpenReaders() ) return false;
275 // "pop" next alignment and reader
276 ReaderAlignment nextReaderAlignment = m_alignments->TakeFirst();
277 BamReader* reader = nextReaderAlignment.first;
278 BamAlignment* alignment = nextReaderAlignment.second;
280 // save it by copy to our argument
281 al = BamAlignment(*alignment);
283 // peek to next alignment & store in cache
284 m_isCoreMode = coreMode;
285 SaveNextAlignment(reader,alignment);
292 bool BamMultiReaderPrivate::Open(const vector<string>& filenames,
295 bool preferStandardIndex)
297 // store core mode flag
298 m_isCoreMode = coreMode;
300 // first clear out any prior alignment cache prior data
301 if ( m_alignments ) {
302 m_alignments->Clear();
307 // create alignment cache based on sorting mode
308 if ( m_isSortedByPosition )
309 m_alignments = new PositionMultiMerger;
311 m_alignments = new ReadNameMultiMerger;
313 // iterate over filenames
314 vector<string>::const_iterator filenameIter = filenames.begin();
315 vector<string>::const_iterator filenameEnd = filenames.end();
316 for ( ; filenameIter != filenameEnd; ++filenameIter ) {
317 const string filename = (*filenameIter);
319 bool openedOk = true;
320 BamReader* reader = new BamReader;
321 openedOk = reader->Open(filename, "", openIndexes, preferStandardIndex);
326 // try to read first alignment
328 BamAlignment* alignment = new BamAlignment;
329 fileOk &= ( coreMode ? reader->GetNextAlignmentCore(*alignment)
330 : reader->GetNextAlignment(*alignment) );
334 m_readers.push_back( make_pair(reader, alignment) );
335 m_alignments->Add( make_pair(reader, alignment) );
338 cerr << "WARNING: could not read first alignment in "
339 << filename << ", ignoring file" << endl;
341 // if only file available & could not be read, return failure
342 if ( filenames.size() == 1 )
347 // TODO; any further error handling when openedOK is false ??
351 // files opened ok, at least one alignment could be read,
352 // now need to check that all files use same reference data
357 // print associated filenames to stdout
358 void BamMultiReaderPrivate::PrintFilenames(void) const {
360 vector<ReaderAlignment>::const_iterator readerIter = m_readers.begin();
361 vector<ReaderAlignment>::const_iterator readerEnd = m_readers.end();
362 for ( ; readerIter != readerEnd; ++readerIter ) {
363 const BamReader* reader = (*readerIter).first;
364 if ( reader == 0 ) continue;
365 cout << reader->GetFilename() << endl;
369 // returns BAM file pointers to beginning of alignment data
370 bool BamMultiReaderPrivate::Rewind(void) {
373 vector<ReaderAlignment>::iterator readerIter = m_readers.begin();
374 vector<ReaderAlignment>::iterator readerEnd = m_readers.end();
375 for ( ; readerIter != readerEnd; ++readerIter ) {
376 BamReader* reader = (*readerIter).first;
377 if ( reader == 0 ) continue;
378 result &= reader->Rewind();
383 void BamMultiReaderPrivate::SaveNextAlignment(BamReader* reader, BamAlignment* alignment) {
385 // must be in core mode && sorting by position to call GNACore()
386 if ( m_isCoreMode && m_isSortedByPosition ) {
387 if ( reader->GetNextAlignmentCore(*alignment) )
388 m_alignments->Add( make_pair(reader, alignment) );
391 // not in core mode and/or sorting by readname, must call GNA()
393 if ( reader->GetNextAlignment(*alignment) )
394 m_alignments->Add( make_pair(reader, alignment) );
398 // sets the index caching mode on the readers
399 void BamMultiReaderPrivate::SetIndexCacheMode(const BamIndex::BamIndexCacheMode mode) {
401 vector<ReaderAlignment>::iterator readerIter = m_readers.begin();
402 vector<ReaderAlignment>::iterator readerEnd = m_readers.end();
403 for ( ; readerIter != readerEnd; ++readerIter ) {
404 BamReader* reader = (*readerIter).first;
405 if ( reader == 0 ) continue;
406 reader->SetIndexCacheMode(mode);
410 bool BamMultiReaderPrivate::SetRegion(const BamRegion& region) {
412 // NB: While it may make sense to track readers in which we can
413 // successfully SetRegion, In practice a failure of SetRegion means "no
414 // alignments here." It makes sense to simply accept the failure,
415 // UpdateAlignments(), and continue.
417 vector<ReaderAlignment>::iterator readerIter = m_readers.begin();
418 vector<ReaderAlignment>::iterator readerEnd = m_readers.end();
419 for ( ; readerIter != readerEnd; ++readerIter ) {
420 BamReader* reader = (*readerIter).first;
421 if ( reader == 0 ) continue;
422 if ( !reader->SetRegion(region) ) {
423 cerr << "ERROR: could not jump " << reader->GetFilename() << " to "
424 << region.LeftRefID << ":" << region.LeftPosition << ".."
425 << region.RightRefID << ":" << region.RightPosition << endl;
433 void BamMultiReaderPrivate::SetSortOrder(const BamMultiReader::SortOrder& order) {
435 const BamMultiReader::SortOrder oldSortOrder = ( m_isSortedByPosition ? BamMultiReader::SortedByPosition
436 : BamMultiReader::SortedByReadName ) ;
437 // skip if no change needed
438 if ( oldSortOrder == order ) return;
440 // create new alignment cache
441 IBamMultiMerger* newAlignmentCache(0);
442 if ( order == BamMultiReader::SortedByPosition ) {
443 newAlignmentCache = new PositionMultiMerger;
444 m_isSortedByPosition = true;
447 newAlignmentCache = new ReadNameMultiMerger;
448 m_isSortedByPosition = false;
451 // copy old cache contents to new cache
452 while ( m_alignments->Size() > 0 ) {
453 ReaderAlignment value = m_alignments->TakeFirst();
454 newAlignmentCache->Add(value);
457 // remove old cache structure & point to new cache
459 m_alignments = newAlignmentCache;
462 // updates our alignment cache
463 void BamMultiReaderPrivate::UpdateAlignments(void) {
466 m_alignments->Clear();
468 // iterate over readers
469 vector<ReaderAlignment>::iterator readerIter = m_readers.begin();
470 vector<ReaderAlignment>::iterator readerEnd = m_readers.end();
471 for ( ; readerIter != readerEnd; ++readerIter ) {
472 BamReader* reader = (*readerIter).first;
473 BamAlignment* alignment = (*readerIter).second;
474 if ( reader == 0 ) continue;
475 SaveNextAlignment(reader, alignment);
479 // splits the entire header into a list of strings
480 const vector<string> BamMultiReaderPrivate::SplitHeaderText(const string& headerText) const {
481 stringstream header(headerText);
482 vector<string> lines;
484 while ( getline(header, item) )
485 lines.push_back(item);
489 // ValidateReaders checks that all the readers point to BAM files representing
490 // alignments against the same set of reference sequences, and that the
491 // sequences are identically ordered. If these checks fail the operation of
492 // the multireader is undefined, so we force program exit.
493 void BamMultiReaderPrivate::ValidateReaders(void) const {
495 // retrieve first reader data
496 const BamReader* firstReader = m_readers.front().first;
497 if ( firstReader == 0 ) return; // signal error?
498 const RefVector firstReaderRefData = firstReader->GetReferenceData();
499 const int firstReaderRefCount = firstReader->GetReferenceCount();
500 const int firstReaderRefSize = firstReaderRefData.size();
502 // iterate over all readers
503 vector<ReaderAlignment>::const_iterator readerIter = m_readers.begin();
504 vector<ReaderAlignment>::const_iterator readerEnd = m_readers.end();
505 for ( ; readerIter != readerEnd; ++readerIter ) {
507 // get current reader data
508 BamReader* reader = (*readerIter).first;
509 if ( reader == 0 ) continue; // error?
510 const RefVector currentReaderRefData = reader->GetReferenceData();
511 const int currentReaderRefCount = reader->GetReferenceCount();
512 const int currentReaderRefSize = currentReaderRefData.size();
514 // init container iterators
515 RefVector::const_iterator firstRefIter = firstReaderRefData.begin();
516 RefVector::const_iterator firstRefEnd = firstReaderRefData.end();
517 RefVector::const_iterator currentRefIter = currentReaderRefData.begin();
519 // compare reference counts from BamReader ( & container size, in case of BR error)
520 if ( (currentReaderRefCount != firstReaderRefCount) ||
521 (firstReaderRefSize != currentReaderRefSize) )
523 cerr << "ERROR: mismatched number of references in " << reader->GetFilename()
524 << " expected " << firstReaderRefCount
525 << " reference sequences but only found " << currentReaderRefCount << endl;
529 // this will be ok; we just checked above that we have identically-sized sets of references
530 // here we simply check if they are all, in fact, equal in content
531 while ( firstRefIter != firstRefEnd ) {
533 const RefData& firstRef = (*firstRefIter);
534 const RefData& currentRef = (*currentRefIter);
536 // compare reference name & length
537 if ( (firstRef.RefName != currentRef.RefName) ||
538 (firstRef.RefLength != currentRef.RefLength) )
540 cerr << "ERROR: mismatched references found in " << reader->GetFilename()
541 << " expected: " << endl;
542 RefVector::const_iterator refIter = firstReaderRefData.begin();
543 RefVector::const_iterator refEnd = firstReaderRefData.end();
544 for ( ; refIter != refEnd; ++refIter ) {
545 const RefData& entry = (*refIter);
546 cerr << entry.RefName << " " << entry.RefLength << endl;
549 cerr << "but found: " << endl;
550 refIter = currentReaderRefData.begin();
551 refEnd = currentReaderRefData.end();
552 for ( ; refIter != refEnd; ++refIter ) {
553 const RefData& entry = (*refIter);
554 cerr << entry.RefName << " " << entry.RefLength << endl;