1 // ***************************************************************************
2 // BamStandardIndex.cpp (c) 2010 Derek Barnett
3 // Marth Lab, Department of Biology, Boston College
4 // All rights reserved.
5 // ---------------------------------------------------------------------------
6 // Last modified: 5 April 2011 (DB)
7 // ---------------------------------------------------------------------------
8 // Provides index operations for the standardized BAM index format (".bai")
9 // ***************************************************************************
11 #include <api/BamAlignment.h>
12 #include <api/internal/BamReader_p.h>
13 #include <api/internal/BamStandardIndex_p.h>
14 using namespace BamTools;
15 using namespace BamTools::Internal;
24 // static BamStandardIndex constants
25 const int BamStandardIndex::MAX_BIN = 37450; // =(8^6-1)/7+1
26 const int BamStandardIndex::BAM_LIDX_SHIFT = 14;
27 const string BamStandardIndex::BAI_EXTENSION = ".bai";
28 const char* const BamStandardIndex::BAI_MAGIC = "BAI\1";
29 const int BamStandardIndex::SIZEOF_ALIGNMENTCHUNK = sizeof(uint64_t)*2;
30 const int BamStandardIndex::SIZEOF_BINCORE = sizeof(uint32_t) + sizeof(int32_t);
31 const int BamStandardIndex::SIZEOF_LINEAROFFSET = sizeof(uint64_t);
34 BamStandardIndex::BamStandardIndex(Internal::BamReaderPrivate* reader)
37 , m_cacheMode(BamIndex::LimitedIndexCaching)
41 m_isBigEndian = BamTools::SystemIsBigEndian();
45 BamStandardIndex::~BamStandardIndex(void) {
49 bool BamStandardIndex::AdjustRegion(const BamRegion& region, uint32_t& begin, uint32_t& end) {
51 // retrieve references from reader
52 const RefVector& references = m_reader->GetReferenceData();
54 // make sure left-bound position is valid
55 if ( region.LeftPosition > references.at(region.LeftRefID).RefLength )
59 begin = (unsigned int)region.LeftPosition;
61 // if right bound specified AND left&right bounds are on same reference
62 // OK to use right bound position as region 'end'
63 if ( region.isRightBoundSpecified() && ( region.LeftRefID == region.RightRefID ) )
64 end = (unsigned int)region.RightPosition;
66 // otherwise, set region 'end' to last reference base
67 else end = (unsigned int)references.at(region.LeftRefID).RefLength - 1;
73 void BamStandardIndex::CalculateCandidateBins(const uint32_t& begin,
75 set<uint16_t>& candidateBins)
77 // initialize list, bin '0' is always a valid bin
78 candidateBins.insert(0);
80 // get rest of bins that contain this region
82 for (k = 1 + (begin>>26); k <= 1 + (end>>26); ++k) { candidateBins.insert(k); }
83 for (k = 9 + (begin>>23); k <= 9 + (end>>23); ++k) { candidateBins.insert(k); }
84 for (k = 73 + (begin>>20); k <= 73 + (end>>20); ++k) { candidateBins.insert(k); }
85 for (k = 585 + (begin>>17); k <= 585 + (end>>17); ++k) { candidateBins.insert(k); }
86 for (k = 4681 + (begin>>14); k <= 4681 + (end>>14); ++k) { candidateBins.insert(k); }
89 bool BamStandardIndex::CalculateCandidateOffsets(const BaiReferenceSummary& refSummary,
90 const uint64_t& minOffset,
91 set<uint16_t>& candidateBins,
92 vector<int64_t>& offsets)
94 // attempt seek to first bin
95 if ( !Seek(refSummary.FirstBinFilePosition, SEEK_SET) )
98 // iterate over reference bins
100 int32_t numAlignmentChunks;
101 set<uint16_t>::iterator candidateBinIter;
102 for ( int i = 0; i < refSummary.NumBins; ++i ) {
104 // read bin contents (if successful, alignment chunks are now in m_buffer)
105 if ( !ReadBinIntoBuffer(binId, numAlignmentChunks) )
108 // see if bin is a 'candidate bin'
109 candidateBinIter = candidateBins.find(binId);
111 // if not, move on to next bin
112 if ( candidateBinIter == candidateBins.end() )
115 // otherwise, check bin's contents against for overlap
118 unsigned int offset = 0;
122 // iterate over alignment chunks
123 for (int j = 0; j < numAlignmentChunks; ++j ) {
125 // read chunk start & stop from buffer
126 memcpy((char*)&chunkStart, m_buffer+offset, sizeof(uint64_t));
127 offset += sizeof(uint64_t);
128 memcpy((char*)&chunkStop, m_buffer, sizeof(uint64_t));
129 offset += sizeof(uint64_t);
131 // swap endian-ness if necessary
132 if ( m_isBigEndian ) {
133 SwapEndian_64(chunkStart);
134 SwapEndian_64(chunkStop);
137 // store alignment chunk's start offset
138 // if its stop offset is larger than our 'minOffset'
139 if ( chunkStop > minOffset )
140 offsets.push_back(chunkStart);
143 // 'pop' bin ID from candidate bins set
144 candidateBins.erase(candidateBinIter);
146 // quit if no more candidates
147 if ( candidateBins.empty() )
152 // return success/failure on calculating at least 1 offset
153 return ( !offsets.empty() );
156 uint64_t BamStandardIndex::CalculateMinOffset(const BaiReferenceSummary& refSummary,
157 const uint32_t& begin)
159 // if no linear offsets exist, return 0
160 if ( refSummary.NumLinearOffsets == 0 )
163 // if 'begin' starts beyond last linear offset, use the last linear offset as minimum
164 // else use the offset corresponding to the requested start position
165 const int shiftedBegin = begin>>BamStandardIndex::BAM_LIDX_SHIFT;
166 if ( shiftedBegin >= refSummary.NumLinearOffsets )
167 return LookupLinearOffset( refSummary, refSummary.NumLinearOffsets-1 );
169 return LookupLinearOffset( refSummary, shiftedBegin );
172 void BamStandardIndex::CheckBufferSize(char*& buffer,
173 unsigned int& bufferLength,
174 const unsigned int& requestedBytes)
177 if ( requestedBytes > bufferLength ) {
178 bufferLength = requestedBytes + 10;
180 buffer = new char[bufferLength];
182 } catch ( std::bad_alloc ) {
183 cerr << "BamStandardIndex ERROR: out of memory when allocating "
184 << requestedBytes << " byes" << endl;
189 void BamStandardIndex::CheckBufferSize(unsigned char*& buffer,
190 unsigned int& bufferLength,
191 const unsigned int& requestedBytes)
194 if ( requestedBytes > bufferLength ) {
195 bufferLength = requestedBytes + 10;
197 buffer = new unsigned char[bufferLength];
199 } catch ( std::bad_alloc ) {
200 cerr << "BamStandardIndex ERROR: out of memory when allocating "
201 << requestedBytes << " byes" << endl;
206 bool BamStandardIndex::CheckMagicNumber(void) {
208 // check 'magic number' to see if file is BAI index
210 size_t elementsRead = fread(magic, sizeof(char), 4, m_indexStream);
211 if ( elementsRead != 4 ) {
212 cerr << "BamStandardIndex ERROR: could not read format 'magic number'" << endl;
216 // compare to expected value
217 if ( strncmp(magic, BamStandardIndex::BAI_MAGIC, 4) != 0 ) {
218 cerr << "BamStandardIndex ERROR: invalid format" << endl;
226 void BamStandardIndex::ClearReferenceEntry(BaiReferenceEntry& refEntry) {
228 refEntry.Bins.clear();
229 refEntry.LinearOffsets.clear();
232 void BamStandardIndex::CloseFile(void) {
236 fclose(m_indexStream);
238 // clear index file summary data
239 m_indexFileSummary.clear();
241 // clean up I/O buffer
247 // builds index from associated BAM file & writes out to index file
248 bool BamStandardIndex::Create(void) {
250 // return false if BamReader is invalid or not open
251 if ( m_reader == 0 || !m_reader->IsOpen() ) {
252 cerr << "BamStandardIndex ERROR: BamReader is not open"
253 << ", aborting index creation" << endl;
258 if ( !m_reader->Rewind() ) {
259 cerr << "BamStandardIndex ERROR: could not rewind BamReader to create index"
260 << ", aborting index creation" << endl;
264 // open new index file (read & write)
265 string indexFilename = m_reader->Filename() + Extension();
266 if ( !OpenFile(indexFilename, "w+b") ) {
267 cerr << "BamStandardIndex ERROR: could not open ouput index file: " << indexFilename
268 << ", aborting index creation" << endl;
272 // initialize BaiFileSummary with number of references
273 const int& numReferences = m_reader->GetReferenceCount();
274 ReserveForSummary(numReferences);
276 // initialize output file
277 bool createdOk = true;
278 createdOk &= WriteHeader();
280 // set up bin, ID, offset, & coordinate markers
281 const uint32_t defaultValue = 0xffffffffu;
282 uint32_t currentBin = defaultValue;
283 uint32_t lastBin = defaultValue;
284 int32_t currentRefID = defaultValue;
285 int32_t lastRefID = defaultValue;
286 uint64_t currentOffset = (uint64_t)m_reader->Tell();
287 uint64_t lastOffset = currentOffset;
288 int32_t lastPosition = defaultValue;
290 // iterate through alignments in BAM file
292 BaiReferenceEntry refEntry;
293 while ( m_reader->LoadNextAlignment(al) ) {
295 // changed to new reference
296 if ( lastRefID != al.RefID ) {
298 // if not first reference, save previous reference data
299 if ( lastRefID != (int32_t)defaultValue ) {
301 SaveAlignmentChunkToBin(refEntry.Bins, currentBin, currentOffset, lastOffset);
302 createdOk &= WriteReferenceEntry(refEntry);
303 ClearReferenceEntry(refEntry);
305 // update bin markers
306 currentOffset = lastOffset;
309 currentRefID = al.RefID;
312 // update reference markers
313 refEntry.ID = al.RefID;
314 lastRefID = al.RefID;
315 lastBin = defaultValue;
318 // if lastPosition greater than current alignment position - file not sorted properly
319 else if ( lastPosition > al.Position ) {
320 cerr << "BamStandardIndex ERROR: BAM file is not properly sorted by coordinate"
321 << ", aborting index creation"
323 << "At alignment: " << al.Name
324 << " : previous position " << lastPosition
325 << " > this alignment position " << al.Position
326 << " on reference id: " << al.RefID << endl;
330 // if alignment's ref ID is valid & its bin is not a 'leaf'
331 if ( (al.RefID >= 0) && (al.Bin < 4681) )
332 SaveLinearOffsetEntry(refEntry.LinearOffsets, al.Position, al.GetEndPosition(), lastOffset);
334 // changed to new BAI bin
335 if ( al.Bin != lastBin ) {
337 // if not first bin on reference, save previous bin data
338 if ( currentBin != defaultValue )
339 SaveAlignmentChunkToBin(refEntry.Bins, currentBin, currentOffset, lastOffset);
342 currentOffset = lastOffset;
345 currentRefID = al.RefID;
347 // if invalid RefID, break out
348 if ( currentRefID < 0 )
352 // make sure that current file pointer is beyond lastOffset
353 if ( m_reader->Tell() <= (int64_t)lastOffset ) {
354 cerr << "BamStandardIndex ERROR: calculating offsets failed"
355 << ", aborting index creation" << endl;
359 // update lastOffset & lastPosition
360 lastOffset = m_reader->Tell();
361 lastPosition = al.Position;
364 // store last alignment chunk to its bin, then write last reference entry
365 if ( currentRefID >= 0 ) {
366 SaveAlignmentChunkToBin(refEntry.Bins, currentBin, currentOffset, lastOffset);
367 createdOk &= WriteReferenceEntry(refEntry);
370 // rewind reader now that we're done building
371 createdOk &= m_reader->Rewind();
377 // returns format's file extension
378 const string BamStandardIndex::Extension(void) {
379 return BamStandardIndex::BAI_EXTENSION;
382 bool BamStandardIndex::GetOffsets(const BamRegion& region, vector<int64_t>& offsets) {
384 // cannot calculate offsets if unknown/invalid reference ID requested
385 if ( region.LeftRefID < 0 || region.LeftRefID >= (int)m_indexFileSummary.size() )
388 // retrieve index summary for left bound reference
389 const BaiReferenceSummary& refSummary = m_indexFileSummary.at(region.LeftRefID);
391 // set up region boundaries based on actual BamReader data
394 if ( !AdjustRegion(region, begin, end) )
397 // retrieve all candidate bin IDs for region
398 set<uint16_t> candidateBins;
399 CalculateCandidateBins(begin, end, candidateBins);
401 // use reference's linear offsets to calculate the minimum offset
402 // that must be considered to find overlap
403 const uint64_t& minOffset = CalculateMinOffset(refSummary, begin);
405 // attempt to use reference summary, minOffset, & candidateBins to calculate offsets
406 if ( !CalculateCandidateOffsets(refSummary, minOffset, candidateBins, offsets) )
409 // ensure that offsets are sorted before returning
410 sort( offsets.begin(), offsets.end() );
416 // returns whether reference has alignments or no
417 bool BamStandardIndex::HasAlignments(const int& referenceID) const {
418 if ( referenceID < 0 || referenceID >= (int)m_indexFileSummary.size() )
420 const BaiReferenceSummary& refSummary = m_indexFileSummary.at(referenceID);
421 return ( refSummary.NumBins > 0 );
424 bool BamStandardIndex::IsFileOpen(void) const {
425 return ( m_indexStream != 0 );
428 // attempts to use index data to jump to @region, returns success/fail
429 // a "successful" jump indicates no error, but not whether this region has data
430 // * thus, the method sets a flag to indicate whether there are alignments
431 // available after the jump position
432 bool BamStandardIndex::Jump(const BamRegion& region, bool* hasAlignmentsInRegion) {
434 // skip if reader is not valid or is not open
435 if ( m_reader == 0 || !m_reader->IsOpen() )
438 // calculate offsets for this region
439 // if failed, print message, set flag, and return failure
440 vector<int64_t> offsets;
441 if ( !GetOffsets(region, offsets) ) {
442 cerr << "BamStandardIndex ERROR: could not jump"
443 << ", unable to retrieve offsets for region" << endl;
444 *hasAlignmentsInRegion = false;
448 // if no offsets retrieved, set flag
449 if ( offsets.empty() )
450 *hasAlignmentsInRegion = false;
452 // iterate through candidate offsets
455 vector<int64_t>::const_iterator offsetIter = offsets.begin();
456 vector<int64_t>::const_iterator offsetEnd = offsets.end();
457 for ( ; offsetIter != offsetEnd; ++offsetIter) {
459 // attempt seek & load first available alignment
460 // set flag to true if data exists
461 result &= m_reader->Seek(*offsetIter);
462 *hasAlignmentsInRegion = m_reader->LoadNextAlignment(al);
464 // if this alignment corresponds to desired position
465 // return success of seeking back to the offset before the 'current offset' (to cover overlaps)
466 if ( ((al.RefID == region.LeftRefID) &&
467 ((al.Position + al.Length) > region.LeftPosition)) ||
468 (al.RefID > region.LeftRefID) )
470 if ( offsetIter != offsets.begin() )
472 return m_reader->Seek(*offsetIter);
476 // if error in jumping, print message & set flag
478 cerr << "BamStandardIndex ERROR: could not jump"
479 << ", there was a problem seeking in BAM file" << endl;
480 *hasAlignmentsInRegion = false;
483 // return success/failure
487 // loads existing data from file into memory
488 bool BamStandardIndex::Load(const std::string& filename) {
490 // attempt open index file (read-only)
491 if ( !OpenFile(filename, "rb") ) {
492 cerr << "BamStandardIndex ERROR: could not open input index file: " << filename
493 << ", aborting index load" << endl;
497 // if invalid format 'magic number', close & return failure
498 if ( !CheckMagicNumber() ) {
503 // attempt to load index file summary, return success/failure
504 return SummarizeIndexFile();
507 uint64_t BamStandardIndex::LookupLinearOffset(const BaiReferenceSummary& refSummary, const int& index) {
509 // attempt seek to proper index file position
510 const int64_t linearOffsetFilePosition = (int64_t)refSummary.FirstLinearOffsetFilePosition +
511 index*BamStandardIndex::SIZEOF_LINEAROFFSET;
512 if ( !Seek(linearOffsetFilePosition, SEEK_SET) )
515 // read linear offset from BAI file
516 uint64_t linearOffset(0);
517 if ( !ReadLinearOffset(linearOffset) )
522 void BamStandardIndex::MergeAlignmentChunks(BaiAlignmentChunkVector& chunks) {
524 // skip if chunks are empty, nothing to merge
525 if ( chunks.empty() )
528 // set up merged alignment chunk container
529 BaiAlignmentChunkVector mergedChunks;
530 mergedChunks.push_back( chunks[0] );
532 // iterate over chunks
534 BaiAlignmentChunkVector::iterator chunkIter = chunks.begin();
535 BaiAlignmentChunkVector::iterator chunkEnd = chunks.end();
536 for ( ++chunkIter; chunkIter != chunkEnd; ++chunkIter) {
538 // get 'currentMergeChunk' based on numeric index
539 BaiAlignmentChunk& currentMergeChunk = mergedChunks[i];
541 // get sourceChunk based on source vector iterator
542 BaiAlignmentChunk& sourceChunk = (*chunkIter);
544 // if currentMergeChunk ends where sourceChunk starts, then merge the two
545 if ( currentMergeChunk.Stop>>16 == sourceChunk.Start>>16 )
546 currentMergeChunk.Stop = sourceChunk.Stop;
550 // append sourceChunk after currentMergeChunk
551 mergedChunks.push_back(sourceChunk);
553 // update i, so the next iteration will consider the
554 // recently-appended sourceChunk as new mergeChunk candidate
559 // saved newly-merged chunks into (parameter) chunks
560 chunks = mergedChunks;
563 bool BamStandardIndex::OpenFile(const std::string& filename, const char* mode) {
565 // make sure any previous index file is closed
568 // attempt to open file
569 m_indexStream = fopen(filename.c_str(), mode);
573 bool BamStandardIndex::ReadBinID(uint32_t& binId) {
574 size_t elementsRead = 0;
575 elementsRead += fread(&binId, sizeof(binId), 1, m_indexStream);
576 if ( m_isBigEndian ) SwapEndian_32(binId);
577 return ( elementsRead == 1 );
580 bool BamStandardIndex::ReadBinIntoBuffer(uint32_t& binId, int32_t& numAlignmentChunks) {
585 readOk &= ReadBinID(binId);
586 readOk &= ReadNumAlignmentChunks(numAlignmentChunks);
589 const unsigned int bytesRequested = numAlignmentChunks*BamStandardIndex::SIZEOF_ALIGNMENTCHUNK;
590 readOk &= ReadIntoBuffer(bytesRequested);
592 // return success/failure
596 bool BamStandardIndex::ReadIntoBuffer(const unsigned int& bytesRequested) {
598 // ensure that our buffer is big enough for request
599 BamStandardIndex::CheckBufferSize(m_buffer, m_bufferLength, bytesRequested);
601 // read from BAI file stream
602 size_t bytesRead = fread( m_buffer, sizeof(char), bytesRequested, m_indexStream );
603 return ( bytesRead == (size_t)bytesRequested );
606 bool BamStandardIndex::ReadLinearOffset(uint64_t& linearOffset) {
607 size_t elementsRead = 0;
608 elementsRead += fread(&linearOffset, sizeof(linearOffset), 1, m_indexStream);
609 if ( m_isBigEndian ) SwapEndian_64(linearOffset);
610 return ( elementsRead == 1 );
613 bool BamStandardIndex::ReadNumAlignmentChunks(int& numAlignmentChunks) {
614 size_t elementsRead = 0;
615 elementsRead += fread(&numAlignmentChunks, sizeof(numAlignmentChunks), 1, m_indexStream);
616 if ( m_isBigEndian ) SwapEndian_32(numAlignmentChunks);
617 return ( elementsRead == 1 );
620 bool BamStandardIndex::ReadNumBins(int& numBins) {
621 size_t elementsRead = 0;
622 elementsRead += fread(&numBins, sizeof(numBins), 1, m_indexStream);
623 if ( m_isBigEndian ) SwapEndian_32(numBins);
624 return ( elementsRead == 1 );
627 bool BamStandardIndex::ReadNumLinearOffsets(int& numLinearOffsets) {
628 size_t elementsRead = 0;
629 elementsRead += fread(&numLinearOffsets, sizeof(numLinearOffsets), 1, m_indexStream);
630 if ( m_isBigEndian ) SwapEndian_32(numLinearOffsets);
631 return ( elementsRead == 1 );
634 bool BamStandardIndex::ReadNumReferences(int& numReferences) {
635 size_t elementsRead = 0;
636 elementsRead += fread(&numReferences, sizeof(numReferences), 1, m_indexStream);
637 if ( m_isBigEndian ) SwapEndian_32(numReferences);
638 return ( elementsRead == 1 );
641 void BamStandardIndex::ReserveForSummary(const int& numReferences) {
642 m_indexFileSummary.clear();
643 m_indexFileSummary.assign( numReferences, BaiReferenceSummary() );
646 void BamStandardIndex::SaveAlignmentChunkToBin(BaiBinMap& binMap,
647 const uint32_t& currentBin,
648 const uint64_t& currentOffset,
649 const uint64_t& lastOffset)
651 // create new alignment chunk
652 BaiAlignmentChunk newChunk(currentOffset, lastOffset);
654 // if no entry exists yet for this bin, create one and store alignment chunk
655 BaiBinMap::iterator binIter = binMap.find(currentBin);
656 if ( binIter == binMap.end() ) {
657 BaiAlignmentChunkVector newChunks;
658 newChunks.push_back(newChunk);
659 binMap.insert( pair<uint32_t, BaiAlignmentChunkVector>(currentBin, newChunks));
662 // otherwise, just append alignment chunk
664 BaiAlignmentChunkVector& binChunks = (*binIter).second;
665 binChunks.push_back( newChunk );
669 void BamStandardIndex::SaveBinsSummary(const int& refId, const int& numBins) {
670 BaiReferenceSummary& refSummary = m_indexFileSummary.at(refId);
671 refSummary.NumBins = numBins;
672 refSummary.FirstBinFilePosition = Tell();
675 void BamStandardIndex::SaveLinearOffsetEntry(BaiLinearOffsetVector& offsets,
676 const int& alignmentStartPosition,
677 const int& alignmentStopPosition,
678 const uint64_t& lastOffset)
680 // get converted offsets
681 const int beginOffset = alignmentStartPosition >> BamStandardIndex::BAM_LIDX_SHIFT;
682 const int endOffset = (alignmentStopPosition - 1) >> BamStandardIndex::BAM_LIDX_SHIFT;
684 // resize vector if necessary
685 int oldSize = offsets.size();
686 int newSize = endOffset + 1;
687 if ( oldSize < newSize )
688 offsets.resize(newSize, 0);
691 for( int i = beginOffset + 1; i <= endOffset; ++i ) {
692 if ( offsets[i] == 0 )
693 offsets[i] = lastOffset;
697 void BamStandardIndex::SaveLinearOffsetsSummary(const int& refId, const int& numLinearOffsets) {
698 BaiReferenceSummary& refSummary = m_indexFileSummary.at(refId);
699 refSummary.NumLinearOffsets = numLinearOffsets;
700 refSummary.FirstLinearOffsetFilePosition = Tell();
703 // seek to position in index file stream
704 bool BamStandardIndex::Seek(const int64_t& position, const int& origin) {
705 return ( fseek64(m_indexStream, position, origin) == 0 );
708 // change the index caching behavior
709 void BamStandardIndex::SetCacheMode(const BamIndex::IndexCacheMode& mode) {
711 // do nothing else here ? cache mode will be ignored from now on, most likely
714 bool BamStandardIndex::SkipBins(const int& numBins) {
716 int32_t numAlignmentChunks;
717 bool skippedOk = true;
718 for (int i = 0; i < numBins; ++i)
719 skippedOk &= ReadBinIntoBuffer(binId, numAlignmentChunks); // results & buffer ignored
723 bool BamStandardIndex::SkipLinearOffsets(const int& numLinearOffsets) {
724 const unsigned int bytesRequested = numLinearOffsets*BamStandardIndex::SIZEOF_LINEAROFFSET;
725 return ReadIntoBuffer(bytesRequested);
728 void BamStandardIndex::SortLinearOffsets(BaiLinearOffsetVector& linearOffsets) {
729 sort( linearOffsets.begin(), linearOffsets.end() );
732 bool BamStandardIndex::SummarizeBins(BaiReferenceSummary& refSummary) {
734 // load number of bins
736 if ( !ReadNumBins(numBins) )
739 // store bins summary for this reference
740 refSummary.NumBins = numBins;
741 refSummary.FirstBinFilePosition = Tell();
743 // attempt skip reference bins, return success/failure
744 return SkipBins(numBins);
747 bool BamStandardIndex::SummarizeIndexFile(void) {
749 // load number of reference sequences
751 if ( !ReadNumReferences(numReferences) )
754 // initialize file summary data
755 ReserveForSummary(numReferences);
757 // iterate over reference entries
758 bool loadedOk = true;
759 BaiFileSummary::iterator summaryIter = m_indexFileSummary.begin();
760 BaiFileSummary::iterator summaryEnd = m_indexFileSummary.end();
761 for ( int i = 0; summaryIter != summaryEnd; ++summaryIter, ++i )
762 loadedOk &= SummarizeReference(*summaryIter);
768 bool BamStandardIndex::SummarizeLinearOffsets(BaiReferenceSummary& refSummary) {
770 // load number of linear offsets
771 int numLinearOffsets;
772 if ( !ReadNumLinearOffsets(numLinearOffsets) )
775 // store bin summary data for this reference
776 refSummary.NumLinearOffsets = numLinearOffsets;
777 refSummary.FirstLinearOffsetFilePosition = Tell();
779 // skip linear offsets in index file
780 return SkipLinearOffsets(numLinearOffsets);
783 bool BamStandardIndex::SummarizeReference(BaiReferenceSummary& refSummary) {
784 bool loadedOk = true;
785 loadedOk &= SummarizeBins(refSummary);
786 loadedOk &= SummarizeLinearOffsets(refSummary);
790 // return position of file pointer in index file stream
791 int64_t BamStandardIndex::Tell(void) const {
792 return ftell64(m_indexStream);
795 bool BamStandardIndex::WriteAlignmentChunk(const BaiAlignmentChunk& chunk) {
797 size_t elementsWritten = 0;
799 // localize alignment chunk offsets
800 uint64_t start = chunk.Start;
801 uint64_t stop = chunk.Stop;
803 // swap endian-ness if necessary
804 if ( m_isBigEndian ) {
805 SwapEndian_64(start);
809 // write to index file
810 elementsWritten += fwrite(&start, sizeof(start), 1, m_indexStream);
811 elementsWritten += fwrite(&stop, sizeof(stop), 1, m_indexStream);
813 // return success/failure of write
814 return ( elementsWritten == 2 );
817 bool BamStandardIndex::WriteAlignmentChunks(BaiAlignmentChunkVector& chunks) {
819 // make sure chunks are merged (simplified) before writing & saving summary
820 MergeAlignmentChunks(chunks);
822 size_t elementsWritten = 0;
825 int32_t chunkCount = chunks.size();
826 if ( m_isBigEndian ) SwapEndian_32(chunkCount);
827 elementsWritten += fwrite(&chunkCount, sizeof(chunkCount), 1, m_indexStream);
829 // iterate over chunks
830 bool chunksOk = true;
831 BaiAlignmentChunkVector::const_iterator chunkIter = chunks.begin();
832 BaiAlignmentChunkVector::const_iterator chunkEnd = chunks.end();
833 for ( ; chunkIter != chunkEnd; ++chunkIter )
834 chunksOk &= WriteAlignmentChunk( (*chunkIter) );
836 // return success/failure of write
837 return ( (elementsWritten == 1) && chunksOk );
840 bool BamStandardIndex::WriteBin(const uint32_t& binId, BaiAlignmentChunkVector& chunks) {
842 size_t elementsWritten = 0;
845 uint32_t binKey = binId;
846 if ( m_isBigEndian ) SwapEndian_32(binKey);
847 elementsWritten += fwrite(&binKey, sizeof(binKey), 1, m_indexStream);
849 // write bin's alignment chunks
850 bool chunksOk = WriteAlignmentChunks(chunks);
852 // return success/failure of write
853 return ( (elementsWritten == 1) && chunksOk );
856 bool BamStandardIndex::WriteBins(const int& refId, BaiBinMap& bins) {
858 size_t elementsWritten = 0;
860 // write number of bins
861 int32_t binCount = bins.size();
862 if ( m_isBigEndian ) SwapEndian_32(binCount);
863 elementsWritten += fwrite(&binCount, sizeof(binCount), 1, m_indexStream);
865 // save summary for reference's bins
866 SaveBinsSummary(refId, bins.size());
870 BaiBinMap::iterator binIter = bins.begin();
871 BaiBinMap::iterator binEnd = bins.end();
872 for ( ; binIter != binEnd; ++binIter )
873 binsOk &= WriteBin( (*binIter).first, (*binIter).second );
875 // return success/failure of write
876 return ( (elementsWritten == 1) && binsOk );
879 bool BamStandardIndex::WriteHeader(void) {
881 size_t elementsWritten = 0;
883 // write magic number
884 elementsWritten += fwrite(BamStandardIndex::BAI_MAGIC, sizeof(char), 4, m_indexStream);
886 // write number of reference sequences
887 int32_t numReferences = m_indexFileSummary.size();
888 if ( m_isBigEndian ) SwapEndian_32(numReferences);
889 elementsWritten += fwrite(&numReferences, sizeof(numReferences), 1, m_indexStream);
891 // return success/failure of write
892 return (elementsWritten == 5);
895 bool BamStandardIndex::WriteLinearOffsets(const int& refId, BaiLinearOffsetVector& linearOffsets) {
897 // make sure linear offsets are sorted before writing & saving summary
898 SortLinearOffsets(linearOffsets);
900 size_t elementsWritten = 0;
902 // write number of linear offsets
903 int32_t offsetCount = linearOffsets.size();
904 if ( m_isBigEndian ) SwapEndian_32(offsetCount);
905 elementsWritten += fwrite(&offsetCount, sizeof(offsetCount), 1, m_indexStream);
907 // save summary for reference's linear offsets
908 SaveLinearOffsetsSummary(refId, linearOffsets.size());
910 // iterate over linear offsets
911 BaiLinearOffsetVector::const_iterator offsetIter = linearOffsets.begin();
912 BaiLinearOffsetVector::const_iterator offsetEnd = linearOffsets.end();
913 for ( ; offsetIter != offsetEnd; ++offsetIter ) {
915 // write linear offset
916 uint64_t linearOffset = (*offsetIter);
917 if ( m_isBigEndian ) SwapEndian_64(linearOffset);
918 elementsWritten += fwrite(&linearOffset, sizeof(linearOffset), 1, m_indexStream);
921 // return success/failure of write
922 return ( elementsWritten == (size_t)(linearOffsets.size() + 1) );
925 bool BamStandardIndex::WriteReferenceEntry(BaiReferenceEntry& refEntry) {
927 refOk &= WriteBins(refEntry.ID, refEntry.Bins);
928 refOk &= WriteLinearOffsets(refEntry.ID, refEntry.LinearOffsets);