1 // ***************************************************************************
\r
2 // BamReader.cpp (c) 2009 Derek Barnett, Michael Str�mberg
\r
3 // Marth Lab, Department of Biology, Boston College
\r
4 // All rights reserved.
\r
5 // ---------------------------------------------------------------------------
\r
6 // Last modified: 14 April 2010 (DB)
\r
7 // ---------------------------------------------------------------------------
\r
8 // Uses BGZF routines were adapted from the bgzf.c code developed at the Broad
\r
10 // ---------------------------------------------------------------------------
\r
11 // Provides the basic functionality for reading BAM files
\r
12 // ***************************************************************************
\r
15 #include <algorithm>
\r
20 // BamTools includes
\r
22 #include "BamReader.h"
\r
23 using namespace BamTools;
\r
24 using namespace std;
\r
26 namespace BamTools {
\r
27 struct BamAlignmentSupportData {
\r
29 uint32_t BlockLength;
\r
30 uint32_t NumCigarOperations;
\r
31 uint32_t QueryNameLength;
\r
32 uint32_t QuerySequenceLength;
\r
34 } // namespace BamTools
\r
36 struct BamReader::BamReaderPrivate {
\r
38 // -------------------------------
\r
40 // -------------------------------
\r
42 // general file data
\r
46 RefVector References;
\r
48 int64_t AlignmentsBeginOffset;
\r
50 string IndexFilename;
\r
55 // user-specified region values
\r
56 bool IsRegionSpecified;
\r
60 // BAM character constants
\r
61 const char* DNA_LOOKUP;
\r
62 const char* CIGAR_LOOKUP;
\r
64 // -------------------------------
\r
65 // constructor & destructor
\r
66 // -------------------------------
\r
67 BamReaderPrivate(void);
\r
68 ~BamReaderPrivate(void);
\r
70 // -------------------------------
\r
71 // "public" interface
\r
72 // -------------------------------
\r
76 bool Jump(int refID, int position = 0);
\r
77 void Open(const string& filename, const string& indexFilename = "");
\r
80 // access alignment data
\r
81 bool GetNextAlignment(BamAlignment& bAlignment);
\r
83 // access auxiliary data
\r
84 int GetReferenceID(const string& refName) const;
\r
87 bool CreateIndex(void);
\r
89 // -------------------------------
\r
91 // -------------------------------
\r
93 // *** reading alignments and auxiliary data *** //
\r
95 // calculate bins that overlap region ( left to reference end for now )
\r
96 int BinsFromRegion(int refID, int left, uint16_t[MAX_BIN]);
\r
97 // fills out character data for BamAlignment data
\r
98 bool BuildCharData(BamAlignment& bAlignment, const BamAlignmentSupportData& supportData);
\r
99 // calculate file offset for first alignment chunk overlapping 'left'
\r
100 int64_t GetOffset(int refID, int left);
\r
101 // checks to see if alignment overlaps current region
\r
102 bool IsOverlap(BamAlignment& bAlignment);
\r
103 // retrieves header text from BAM file
\r
104 void LoadHeaderData(void);
\r
105 // retrieves BAM alignment under file pointer
\r
106 bool LoadNextAlignment(BamAlignment& bAlignment, BamAlignmentSupportData& supportData);
\r
107 // builds reference data structure from BAM file
\r
108 void LoadReferenceData(void);
\r
110 // *** index file handling *** //
\r
112 // calculates index for BAM file
\r
113 bool BuildIndex(void);
\r
114 // clear out inernal index data structure
\r
115 void ClearIndex(void);
\r
116 // saves BAM bin entry for index
\r
117 void InsertBinEntry(BamBinMap& binMap, const uint32_t& saveBin, const uint64_t& saveOffset, const uint64_t& lastOffset);
\r
118 // saves linear offset entry for index
\r
119 void InsertLinearOffset(LinearOffsetVector& offsets, const BamAlignment& bAlignment, const uint64_t& lastOffset);
\r
120 // loads index from BAM index file
\r
121 bool LoadIndex(void);
\r
122 // simplifies index by merging 'chunks'
\r
123 void MergeChunks(void);
\r
124 // saves index to BAM index file
\r
125 bool WriteIndex(void);
\r
128 // -----------------------------------------------------
\r
129 // BamReader implementation (wrapper around BRPrivate)
\r
130 // -----------------------------------------------------
\r
133 BamReader::BamReader(void) {
\r
134 d = new BamReaderPrivate;
\r
138 BamReader::~BamReader(void) {
\r
144 void BamReader::Close(void) { d->Close(); }
\r
145 bool BamReader::Jump(int refID, int position) { return d->Jump(refID, position); }
\r
146 void BamReader::Open(const string& filename, const string& indexFilename) { d->Open(filename, indexFilename); }
\r
147 bool BamReader::Rewind(void) { return d->Rewind(); }
\r
149 // access alignment data
\r
150 bool BamReader::GetNextAlignment(BamAlignment& bAlignment) { return d->GetNextAlignment(bAlignment); }
\r
152 // access auxiliary data
\r
153 const string BamReader::GetHeaderText(void) const { return d->HeaderText; }
\r
154 int BamReader::GetReferenceCount(void) const { return d->References.size(); }
\r
155 const RefVector BamReader::GetReferenceData(void) const { return d->References; }
\r
156 int BamReader::GetReferenceID(const string& refName) const { return d->GetReferenceID(refName); }
\r
158 // index operations
\r
159 bool BamReader::CreateIndex(void) { return d->CreateIndex(); }
\r
161 // -----------------------------------------------------
\r
162 // BamReaderPrivate implementation
\r
163 // -----------------------------------------------------
\r
166 BamReader::BamReaderPrivate::BamReaderPrivate(void)
\r
167 : IsIndexLoaded(false)
\r
168 , AlignmentsBeginOffset(0)
\r
169 , IsRegionSpecified(false)
\r
172 , DNA_LOOKUP("=ACMGRSVTWYHKDBN")
\r
173 , CIGAR_LOOKUP("MIDNSHP")
\r
175 IsBigEndian = SystemIsBigEndian();
\r
179 BamReader::BamReaderPrivate::~BamReaderPrivate(void) {
\r
183 // calculate bins that overlap region ( left to reference end for now )
\r
184 int BamReader::BamReaderPrivate::BinsFromRegion(int refID, int left, uint16_t list[MAX_BIN]) {
\r
186 // get region boundaries
\r
187 uint32_t begin = (unsigned int)left;
\r
188 uint32_t end = (unsigned int)References.at(refID).RefLength - 1;
\r
190 // initialize list, bin '0' always a valid bin
\r
194 // get rest of bins that contain this region
\r
196 for (k = 1 + (begin>>26); k <= 1 + (end>>26); ++k) { list[i++] = k; }
\r
197 for (k = 9 + (begin>>23); k <= 9 + (end>>23); ++k) { list[i++] = k; }
\r
198 for (k = 73 + (begin>>20); k <= 73 + (end>>20); ++k) { list[i++] = k; }
\r
199 for (k = 585 + (begin>>17); k <= 585 + (end>>17); ++k) { list[i++] = k; }
\r
200 for (k = 4681 + (begin>>14); k <= 4681 + (end>>14); ++k) { list[i++] = k; }
\r
202 // return number of bins stored
\r
206 bool BamReader::BamReaderPrivate::BuildCharData(BamAlignment& bAlignment, const BamAlignmentSupportData& supportData) {
\r
208 // calculate character lengths/offsets
\r
209 const unsigned int dataLength = supportData.BlockLength - BAM_CORE_SIZE;
\r
210 const unsigned int seqDataOffset = supportData.QueryNameLength + (supportData.NumCigarOperations * 4);
\r
211 const unsigned int qualDataOffset = seqDataOffset + (supportData.QuerySequenceLength+1)/2;
\r
212 const unsigned int tagDataOffset = qualDataOffset + supportData.QuerySequenceLength;
\r
213 const unsigned int tagDataLength = dataLength - tagDataOffset;
\r
215 // set up char buffers
\r
216 const char* allCharData = supportData.AllCharData.data();
\r
217 const char* seqData = ((const char*)allCharData) + seqDataOffset;
\r
218 const char* qualData = ((const char*)allCharData) + qualDataOffset;
\r
219 char* tagData = ((char*)allCharData) + tagDataOffset;
\r
221 // save query sequence
\r
222 bAlignment.QueryBases.clear();
\r
223 bAlignment.QueryBases.reserve(supportData.QuerySequenceLength);
\r
224 for (unsigned int i = 0; i < supportData.QuerySequenceLength; ++i) {
\r
225 char singleBase = DNA_LOOKUP[ ( ( seqData[(i/2)] >> (4*(1-(i%2)))) & 0xf ) ];
\r
226 bAlignment.QueryBases.append(1, singleBase);
\r
229 // save qualities, converting from numeric QV to 'FASTQ-style' ASCII character
\r
230 bAlignment.Qualities.clear();
\r
231 bAlignment.Qualities.reserve(supportData.QuerySequenceLength);
\r
232 for (unsigned int i = 0; i < supportData.QuerySequenceLength; ++i) {
\r
233 char singleQuality = (char)(qualData[i]+33);
\r
234 bAlignment.Qualities.append(1, singleQuality);
\r
237 // parse CIGAR to build 'AlignedBases'
\r
238 bAlignment.AlignedBases.clear();
\r
239 bAlignment.AlignedBases.reserve(supportData.QuerySequenceLength);
\r
242 vector<CigarOp>::const_iterator cigarIter = bAlignment.CigarData.begin();
\r
243 vector<CigarOp>::const_iterator cigarEnd = bAlignment.CigarData.end();
\r
244 for ( ; cigarIter != cigarEnd; ++cigarIter ) {
\r
246 const CigarOp& op = (*cigarIter);
\r
251 bAlignment.AlignedBases.append(bAlignment.QueryBases.substr(k, op.Length)); // for 'M', 'I' - write bases
\r
255 k += op.Length; // for 'S' - soft clip, skip over query bases
\r
259 bAlignment.AlignedBases.append(op.Length, '-'); // for 'D' - write gap character
\r
263 bAlignment.AlignedBases.append( op.Length, '*' ); // for 'P' - write padding character
\r
267 bAlignment.AlignedBases.append( op.Length, 'N' ); // for 'N' - write N's, skip bases in original query sequence
\r
272 break; // for 'H' - hard clip, do nothing to AlignedBases, move to next op
\r
275 printf("ERROR: Invalid Cigar op type\n"); // shouldn't get here
\r
280 // -----------------------
\r
281 // Added: 3-25-2010 DWB
\r
282 // Fixed: endian-correctness for tag data
\r
283 // -----------------------
\r
284 if ( IsBigEndian ) {
\r
286 while ( (unsigned int)i < tagDataLength ) {
\r
288 i += 2; // skip tag type (e.g. "RG", "NM", etc)
\r
289 uint8_t type = toupper(tagData[i]); // lower & upper case letters have same meaning
\r
290 ++i; // skip value type
\r
300 SwapEndian_16p(&tagData[i]);
\r
301 i+=2; // sizeof(uint16_t)
\r
306 SwapEndian_32p(&tagData[i]);
\r
307 i+=4; // sizeof(uint32_t)
\r
311 SwapEndian_64p(&tagData[i]);
\r
312 i+=8; // sizeof(uint64_t)
\r
317 while (tagData[i]) { ++i; }
\r
318 ++i; // increment one more for null terminator
\r
322 printf("ERROR: Invalid tag value type\n"); // shouldn't get here
\r
329 bAlignment.TagData.clear();
\r
330 bAlignment.TagData.resize(tagDataLength);
\r
331 memcpy((char*)bAlignment.TagData.data(), tagData, tagDataLength);
\r
337 // populates BAM index data structure from BAM file data
\r
338 bool BamReader::BamReaderPrivate::BuildIndex(void) {
\r
340 // check to be sure file is open
\r
341 if (!mBGZF.IsOpen) { return false; }
\r
343 // move file pointer to beginning of alignments
\r
346 // get reference count, reserve index space
\r
347 int numReferences = References.size();
\r
348 for ( int i = 0; i < numReferences; ++i ) {
\r
349 Index.push_back(ReferenceIndex());
\r
352 // sets default constant for bin, ID, offset, coordinate variables
\r
353 const uint32_t defaultValue = 0xffffffffu;
\r
356 uint32_t saveBin(defaultValue);
\r
357 uint32_t lastBin(defaultValue);
\r
359 // reference ID data
\r
360 int32_t saveRefID(defaultValue);
\r
361 int32_t lastRefID(defaultValue);
\r
364 uint64_t saveOffset = mBGZF.Tell();
\r
365 uint64_t lastOffset = saveOffset;
\r
368 int32_t lastCoordinate = defaultValue;
\r
370 BamAlignment bAlignment;
\r
371 while( GetNextAlignment(bAlignment) ) {
\r
373 // change of chromosome, save ID, reset bin
\r
374 if ( lastRefID != bAlignment.RefID ) {
\r
375 lastRefID = bAlignment.RefID;
\r
376 lastBin = defaultValue;
\r
379 // if lastCoordinate greater than BAM position - file not sorted properly
\r
380 else if ( lastCoordinate > bAlignment.Position ) {
\r
381 printf("BAM file not properly sorted:\n");
\r
382 printf("Alignment %s : %d > %d on reference (id = %d)", bAlignment.Name.c_str(), lastCoordinate, bAlignment.Position, bAlignment.RefID);
\r
386 // if valid reference && BAM bin spans some minimum cutoff (smaller bin ids span larger regions)
\r
387 if ( (bAlignment.RefID >= 0) && (bAlignment.Bin < 4681) ) {
\r
389 // save linear offset entry (matched to BAM entry refID)
\r
390 ReferenceIndex& refIndex = Index.at(bAlignment.RefID);
\r
391 LinearOffsetVector& offsets = refIndex.Offsets;
\r
392 InsertLinearOffset(offsets, bAlignment, lastOffset);
\r
395 // if current BamAlignment bin != lastBin, "then possibly write the binning index"
\r
396 if ( bAlignment.Bin != lastBin ) {
\r
398 // if not first time through
\r
399 if ( saveBin != defaultValue ) {
\r
401 // save Bam bin entry
\r
402 ReferenceIndex& refIndex = Index.at(saveRefID);
\r
403 BamBinMap& binMap = refIndex.Bins;
\r
404 InsertBinEntry(binMap, saveBin, saveOffset, lastOffset);
\r
407 // update saveOffset
\r
408 saveOffset = lastOffset;
\r
410 // update bin values
\r
411 saveBin = bAlignment.Bin;
\r
412 lastBin = bAlignment.Bin;
\r
414 // update saveRefID
\r
415 saveRefID = bAlignment.RefID;
\r
417 // if invalid RefID, break out (why?)
\r
418 if ( saveRefID < 0 ) { break; }
\r
421 // make sure that current file pointer is beyond lastOffset
\r
422 if ( mBGZF.Tell() <= (int64_t)lastOffset ) {
\r
423 printf("Error in BGZF offsets.\n");
\r
427 // update lastOffset
\r
428 lastOffset = mBGZF.Tell();
\r
430 // update lastCoordinate
\r
431 lastCoordinate = bAlignment.Position;
\r
434 // save any leftover BAM data (as long as refID is valid)
\r
435 if ( saveRefID >= 0 ) {
\r
436 // save Bam bin entry
\r
437 ReferenceIndex& refIndex = Index.at(saveRefID);
\r
438 BamBinMap& binMap = refIndex.Bins;
\r
439 InsertBinEntry(binMap, saveBin, saveOffset, lastOffset);
\r
442 // simplify index by merging chunks
\r
445 // iterate over references
\r
446 BamIndex::iterator indexIter = Index.begin();
\r
447 BamIndex::iterator indexEnd = Index.end();
\r
448 for ( int i = 0; indexIter != indexEnd; ++indexIter, ++i ) {
\r
450 // get reference index data
\r
451 ReferenceIndex& refIndex = (*indexIter);
\r
452 BamBinMap& binMap = refIndex.Bins;
\r
453 LinearOffsetVector& offsets = refIndex.Offsets;
\r
455 // store whether reference has alignments or no
\r
456 References[i].RefHasAlignments = ( binMap.size() > 0 );
\r
458 // sort linear offsets
\r
459 sort(offsets.begin(), offsets.end());
\r
463 // rewind file pointer to beginning of alignments, return success/fail
\r
468 // clear index data structure
\r
469 void BamReader::BamReaderPrivate::ClearIndex(void) {
\r
470 Index.clear(); // sufficient ??
\r
473 // closes the BAM file
\r
474 void BamReader::BamReaderPrivate::Close(void) {
\r
477 HeaderText.clear();
\r
478 IsRegionSpecified = false;
\r
481 // create BAM index from BAM file (keep structure in memory) and write to default index output file
\r
482 bool BamReader::BamReaderPrivate::CreateIndex(void) {
\r
487 // build (& save) index from BAM file
\r
489 ok &= BuildIndex();
\r
490 ok &= WriteIndex();
\r
492 // return success/fail
\r
496 // returns RefID for given RefName (returns References.size() if not found)
\r
497 int BamReader::BamReaderPrivate::GetReferenceID(const string& refName) const {
\r
499 // retrieve names from reference data
\r
500 vector<string> refNames;
\r
501 RefVector::const_iterator refIter = References.begin();
\r
502 RefVector::const_iterator refEnd = References.end();
\r
503 for ( ; refIter != refEnd; ++refIter) {
\r
504 refNames.push_back( (*refIter).RefName );
\r
507 // return 'index-of' refName ( if not found, returns refNames.size() )
\r
508 return distance(refNames.begin(), find(refNames.begin(), refNames.end(), refName));
\r
511 // get next alignment (from specified region, if given)
\r
512 bool BamReader::BamReaderPrivate::GetNextAlignment(BamAlignment& bAlignment) {
\r
514 BamAlignmentSupportData supportData;
\r
516 // if valid alignment available
\r
517 if ( LoadNextAlignment(bAlignment, supportData) ) {
\r
519 // if region not specified, return success
\r
520 if ( !IsRegionSpecified ) {
\r
521 bool ok = BuildCharData(bAlignment, supportData);
\r
525 // load next alignment until region overlap is found
\r
526 while ( !IsOverlap(bAlignment) ) {
\r
527 // if no valid alignment available (likely EOF) return failure
\r
528 if ( !LoadNextAlignment(bAlignment, supportData) ) { return false; }
\r
531 // return success (alignment found that overlaps region)
\r
532 bool ok = BuildCharData(bAlignment, supportData);
\r
536 // no valid alignment
\r
537 else { return false; }
\r
540 // calculate closest indexed file offset for region specified
\r
541 int64_t BamReader::BamReaderPrivate::GetOffset(int refID, int left) {
\r
543 // calculate which bins overlap this region
\r
544 uint16_t* bins = (uint16_t*)calloc(MAX_BIN, 2);
\r
545 int numBins = BinsFromRegion(refID, left, bins);
\r
547 // get bins for this reference
\r
548 const ReferenceIndex& refIndex = Index.at(refID);
\r
549 const BamBinMap& binMap = refIndex.Bins;
\r
551 // get minimum offset to consider
\r
552 const LinearOffsetVector& offsets = refIndex.Offsets;
\r
553 uint64_t minOffset = ( (unsigned int)(left>>BAM_LIDX_SHIFT) >= offsets.size() ) ? 0 : offsets.at(left>>BAM_LIDX_SHIFT);
\r
555 // store offsets to beginning of alignment 'chunks'
\r
556 std::vector<int64_t> chunkStarts;
\r
558 // store all alignment 'chunk' starts for bins in this region
\r
559 for (int i = 0; i < numBins; ++i ) {
\r
560 uint16_t binKey = bins[i];
\r
562 map<uint32_t, ChunkVector>::const_iterator binIter = binMap.find(binKey);
\r
563 if ( (binIter != binMap.end()) && ((*binIter).first == binKey) ) {
\r
565 const ChunkVector& chunks = (*binIter).second;
\r
566 std::vector<Chunk>::const_iterator chunksIter = chunks.begin();
\r
567 std::vector<Chunk>::const_iterator chunksEnd = chunks.end();
\r
568 for ( ; chunksIter != chunksEnd; ++chunksIter) {
\r
569 const Chunk& chunk = (*chunksIter);
\r
570 if ( chunk.Stop > minOffset ) {
\r
571 chunkStarts.push_back( chunk.Start );
\r
580 // if no alignments found, else return smallest offset for alignment starts
\r
581 if ( chunkStarts.size() == 0 ) { return -1; }
\r
582 else { return *min_element(chunkStarts.begin(), chunkStarts.end()); }
\r
585 // saves BAM bin entry for index
\r
586 void BamReader::BamReaderPrivate::InsertBinEntry(BamBinMap& binMap,
\r
587 const uint32_t& saveBin,
\r
588 const uint64_t& saveOffset,
\r
589 const uint64_t& lastOffset)
\r
592 BamBinMap::iterator binIter = binMap.find(saveBin);
\r
594 // create new chunk
\r
595 Chunk newChunk(saveOffset, lastOffset);
\r
597 // if entry doesn't exist
\r
598 if ( binIter == binMap.end() ) {
\r
599 ChunkVector newChunks;
\r
600 newChunks.push_back(newChunk);
\r
601 binMap.insert( pair<uint32_t, ChunkVector>(saveBin, newChunks));
\r
606 ChunkVector& binChunks = (*binIter).second;
\r
607 binChunks.push_back( newChunk );
\r
611 // saves linear offset entry for index
\r
612 void BamReader::BamReaderPrivate::InsertLinearOffset(LinearOffsetVector& offsets,
\r
613 const BamAlignment& bAlignment,
\r
614 const uint64_t& lastOffset)
\r
616 // get converted offsets
\r
617 int beginOffset = bAlignment.Position >> BAM_LIDX_SHIFT;
618 int endOffset = (bAlignment.GetEndPosition() - 1) >> BAM_LIDX_SHIFT;
\r
620 // resize vector if necessary
\r
621 int oldSize = offsets.size();
\r
622 int newSize = endOffset + 1;
\r
623 if ( oldSize < newSize ) { offsets.resize(newSize, 0); }
\r
626 for(int i = beginOffset + 1; i <= endOffset ; ++i) {
\r
627 if ( offsets[i] == 0) {
\r
628 offsets[i] = lastOffset;
\r
633 // returns whether alignment overlaps currently specified region (refID, leftBound)
\r
634 bool BamReader::BamReaderPrivate::IsOverlap(BamAlignment& bAlignment) {
\r
636 // if on different reference sequence, quit
\r
637 if ( bAlignment.RefID != CurrentRefID ) { return false; }
\r
639 // read starts after left boundary
\r
640 if ( bAlignment.Position >= CurrentLeft) { return true; }
\r
642 // return whether alignment end overlaps left boundary
643 return ( bAlignment.GetEndPosition() >= CurrentLeft );
\r
646 // jumps to specified region(refID, leftBound) in BAM file, returns success/fail
\r
647 bool BamReader::BamReaderPrivate::Jump(int refID, int position) {
\r
649 // if data exists for this reference and position is valid
\r
650 if ( References.at(refID).RefHasAlignments && (position <= References.at(refID).RefLength) ) {
\r
652 // set current region
\r
653 CurrentRefID = refID;
\r
654 CurrentLeft = position;
\r
655 IsRegionSpecified = true;
\r
657 // calculate offset
\r
658 int64_t offset = GetOffset(CurrentRefID, CurrentLeft);
\r
660 // if in valid offset, return failure
\r
661 if ( offset == -1 ) { return false; }
\r
663 // otherwise return success of seek operation
\r
664 else { return mBGZF.Seek(offset); }
\r
667 // invalid jump request parameters, return failure
\r
671 // load BAM header data
\r
672 void BamReader::BamReaderPrivate::LoadHeaderData(void) {
\r
674 // check to see if proper BAM header
\r
676 if (mBGZF.Read(buffer, 4) != 4) {
\r
677 printf("Could not read header type\n");
\r
681 if (strncmp(buffer, "BAM\001", 4)) {
\r
682 printf("wrong header type!\n");
\r
686 // get BAM header text length
\r
687 mBGZF.Read(buffer, 4);
\r
688 unsigned int headerTextLength = BgzfData::UnpackUnsignedInt(buffer);
\r
689 if ( IsBigEndian ) { SwapEndian_32(headerTextLength); }
\r
691 // get BAM header text
\r
692 char* headerText = (char*)calloc(headerTextLength + 1, 1);
\r
693 mBGZF.Read(headerText, headerTextLength);
\r
694 HeaderText = (string)((const char*)headerText);
\r
696 // clean up calloc-ed temp variable
\r
700 // load existing index data from BAM index file (".bai"), return success/fail
\r
701 bool BamReader::BamReaderPrivate::LoadIndex(void) {
\r
703 // clear out index data
\r
706 // skip if index file empty
\r
707 if ( IndexFilename.empty() ) { return false; }
\r
709 // open index file, abort on error
\r
710 FILE* indexStream = fopen(IndexFilename.c_str(), "rb");
\r
712 printf("ERROR: Unable to open the BAM index file %s for reading.\n", IndexFilename.c_str() );
\r
716 size_t elementsRead = 0;
\r
718 // see if index is valid BAM index
\r
720 elementsRead = fread(magic, 1, 4, indexStream);
\r
721 if (strncmp(magic, "BAI\1", 4)) {
\r
722 printf("Problem with index file - invalid format.\n");
\r
723 fclose(indexStream);
\r
727 // get number of reference sequences
\r
728 uint32_t numRefSeqs;
\r
729 elementsRead = fread(&numRefSeqs, 4, 1, indexStream);
\r
730 if ( IsBigEndian ) { SwapEndian_32(numRefSeqs); }
\r
732 // intialize space for BamIndex data structure
\r
733 Index.reserve(numRefSeqs);
\r
735 // iterate over reference sequences
\r
736 for (unsigned int i = 0; i < numRefSeqs; ++i) {
\r
738 // get number of bins for this reference sequence
\r
740 elementsRead = fread(&numBins, 4, 1, indexStream);
\r
741 if ( IsBigEndian ) { SwapEndian_32(numBins); }
\r
744 RefData& refEntry = References[i];
\r
745 refEntry.RefHasAlignments = true;
\r
748 // intialize BinVector
\r
751 // iterate over bins for that reference sequence
\r
752 for (int j = 0; j < numBins; ++j) {
\r
756 elementsRead = fread(&binID, 4, 1, indexStream);
\r
758 // get number of regionChunks in this bin
\r
759 uint32_t numChunks;
\r
760 elementsRead = fread(&numChunks, 4, 1, indexStream);
\r
762 if ( IsBigEndian ) {
\r
763 SwapEndian_32(binID);
\r
764 SwapEndian_32(numChunks);
\r
767 // intialize ChunkVector
\r
768 ChunkVector regionChunks;
\r
769 regionChunks.reserve(numChunks);
\r
771 // iterate over regionChunks in this bin
\r
772 for (unsigned int k = 0; k < numChunks; ++k) {
\r
774 // get chunk boundaries (left, right)
\r
777 elementsRead = fread(&left, 8, 1, indexStream);
\r
778 elementsRead = fread(&right, 8, 1, indexStream);
\r
780 if ( IsBigEndian ) {
\r
781 SwapEndian_64(left);
\r
782 SwapEndian_64(right);
\r
786 regionChunks.push_back( Chunk(left, right) );
\r
789 // sort chunks for this bin
\r
790 sort( regionChunks.begin(), regionChunks.end(), ChunkLessThan );
\r
792 // save binID, chunkVector for this bin
\r
793 binMap.insert( pair<uint32_t, ChunkVector>(binID, regionChunks) );
\r
796 // load linear index for this reference sequence
\r
798 // get number of linear offsets
\r
799 int32_t numLinearOffsets;
\r
800 elementsRead = fread(&numLinearOffsets, 4, 1, indexStream);
\r
801 if ( IsBigEndian ) { SwapEndian_32(numLinearOffsets); }
\r
803 // intialize LinearOffsetVector
\r
804 LinearOffsetVector offsets;
\r
805 offsets.reserve(numLinearOffsets);
\r
807 // iterate over linear offsets for this reference sequeence
\r
808 uint64_t linearOffset;
\r
809 for (int j = 0; j < numLinearOffsets; ++j) {
\r
810 // read a linear offset & store
\r
811 elementsRead = fread(&linearOffset, 8, 1, indexStream);
\r
812 if ( IsBigEndian ) { SwapEndian_64(linearOffset); }
\r
813 offsets.push_back(linearOffset);
\r
816 // sort linear offsets
\r
817 sort( offsets.begin(), offsets.end() );
\r
819 // store index data for that reference sequence
\r
820 Index.push_back( ReferenceIndex(binMap, offsets) );
\r
823 // close index file (.bai) and return
\r
824 fclose(indexStream);
\r
828 // populates BamAlignment with alignment data under file pointer, returns success/fail
\r
829 bool BamReader::BamReaderPrivate::LoadNextAlignment(BamAlignment& bAlignment, BamAlignmentSupportData& supportData) {
\r
831 // read in the 'block length' value, make sure it's not zero
\r
833 mBGZF.Read(buffer, 4);
\r
834 supportData.BlockLength = BgzfData::UnpackUnsignedInt(buffer);
\r
835 if ( IsBigEndian ) { SwapEndian_32(supportData.BlockLength); }
\r
836 if ( supportData.BlockLength == 0 ) { return false; }
\r
838 // read in core alignment data, make sure the right size of data was read
\r
839 char x[BAM_CORE_SIZE];
\r
840 if ( mBGZF.Read(x, BAM_CORE_SIZE) != BAM_CORE_SIZE ) { return false; }
\r
842 if ( IsBigEndian ) {
\r
843 for ( int i = 0; i < BAM_CORE_SIZE; i+=sizeof(uint32_t) ) {
\r
844 SwapEndian_32p(&x[i]);
\r
848 // set BamAlignment 'core' and 'support' data
\r
849 bAlignment.RefID = BgzfData::UnpackSignedInt(&x[0]);
\r
850 bAlignment.Position = BgzfData::UnpackSignedInt(&x[4]);
\r
852 unsigned int tempValue = BgzfData::UnpackUnsignedInt(&x[8]);
\r
853 bAlignment.Bin = tempValue >> 16;
\r
854 bAlignment.MapQuality = tempValue >> 8 & 0xff;
\r
855 supportData.QueryNameLength = tempValue & 0xff;
\r
857 tempValue = BgzfData::UnpackUnsignedInt(&x[12]);
\r
858 bAlignment.AlignmentFlag = tempValue >> 16;
\r
859 supportData.NumCigarOperations = tempValue & 0xffff;
\r
861 supportData.QuerySequenceLength = BgzfData::UnpackUnsignedInt(&x[16]);
\r
862 bAlignment.MateRefID = BgzfData::UnpackSignedInt(&x[20]);
\r
863 bAlignment.MatePosition = BgzfData::UnpackSignedInt(&x[24]);
\r
864 bAlignment.InsertSize = BgzfData::UnpackSignedInt(&x[28]);
\r
866 // store 'all char data' and cigar ops
\r
867 const unsigned int dataLength = supportData.BlockLength - BAM_CORE_SIZE;
\r
868 const unsigned int cigarDataOffset = supportData.QueryNameLength;
\r
870 char* allCharData = (char*)calloc(sizeof(char), dataLength);
\r
871 uint32_t* cigarData = (uint32_t*)(allCharData + cigarDataOffset);
\r
873 // read in character data - make sure proper data size was read
\r
874 if ( mBGZF.Read(allCharData, dataLength) != (signed int)dataLength) { return false; }
\r
877 // store alignment name and length
\r
878 bAlignment.Name.assign((const char*)(allCharData));
\r
879 bAlignment.Length = supportData.QuerySequenceLength;
\r
881 // store remaining 'allCharData' in supportData structure
\r
882 supportData.AllCharData.assign((const char*)allCharData, dataLength);
\r
884 // save CigarOps for BamAlignment
\r
885 bAlignment.CigarData.clear();
\r
886 for (unsigned int i = 0; i < supportData.NumCigarOperations; ++i) {
\r
888 // swap if necessary
\r
889 if ( IsBigEndian ) { SwapEndian_32(cigarData[i]); }
\r
891 // build CigarOp structure
\r
893 op.Length = (cigarData[i] >> BAM_CIGAR_SHIFT);
\r
894 op.Type = CIGAR_LOOKUP[ (cigarData[i] & BAM_CIGAR_MASK) ];
\r
897 bAlignment.CigarData.push_back(op);
\r
905 // loads reference data from BAM file
\r
906 void BamReader::BamReaderPrivate::LoadReferenceData(void) {
\r
908 // get number of reference sequences
\r
910 mBGZF.Read(buffer, 4);
\r
911 unsigned int numberRefSeqs = BgzfData::UnpackUnsignedInt(buffer);
\r
912 if ( IsBigEndian ) { SwapEndian_32(numberRefSeqs); }
\r
913 if (numberRefSeqs == 0) { return; }
\r
914 References.reserve((int)numberRefSeqs);
\r
916 // iterate over all references in header
\r
917 for (unsigned int i = 0; i != numberRefSeqs; ++i) {
\r
919 // get length of reference name
\r
920 mBGZF.Read(buffer, 4);
\r
921 unsigned int refNameLength = BgzfData::UnpackUnsignedInt(buffer);
\r
922 if ( IsBigEndian ) { SwapEndian_32(refNameLength); }
\r
923 char* refName = (char*)calloc(refNameLength, 1);
\r
925 // get reference name and reference sequence length
\r
926 mBGZF.Read(refName, refNameLength);
\r
927 mBGZF.Read(buffer, 4);
\r
928 int refLength = BgzfData::UnpackSignedInt(buffer);
\r
929 if ( IsBigEndian ) { SwapEndian_32(refLength); }
\r
931 // store data for reference
\r
932 RefData aReference;
\r
933 aReference.RefName = (string)((const char*)refName);
\r
934 aReference.RefLength = refLength;
\r
935 References.push_back(aReference);
\r
937 // clean up calloc-ed temp variable
\r
942 // merges 'alignment chunks' in BAM bin (used for index building)
\r
943 void BamReader::BamReaderPrivate::MergeChunks(void) {
\r
945 // iterate over reference enties
\r
946 BamIndex::iterator indexIter = Index.begin();
\r
947 BamIndex::iterator indexEnd = Index.end();
\r
948 for ( ; indexIter != indexEnd; ++indexIter ) {
\r
950 // get BAM bin map for this reference
\r
951 ReferenceIndex& refIndex = (*indexIter);
\r
952 BamBinMap& bamBinMap = refIndex.Bins;
\r
954 // iterate over BAM bins
\r
955 BamBinMap::iterator binIter = bamBinMap.begin();
\r
956 BamBinMap::iterator binEnd = bamBinMap.end();
\r
957 for ( ; binIter != binEnd; ++binIter ) {
\r
959 // get chunk vector for this bin
\r
960 ChunkVector& binChunks = (*binIter).second;
\r
961 if ( binChunks.size() == 0 ) { continue; }
\r
963 ChunkVector mergedChunks;
\r
964 mergedChunks.push_back( binChunks[0] );
\r
966 // iterate over chunks
\r
968 ChunkVector::iterator chunkIter = binChunks.begin();
\r
969 ChunkVector::iterator chunkEnd = binChunks.end();
\r
970 for ( ++chunkIter; chunkIter != chunkEnd; ++chunkIter) {
\r
972 // get 'currentChunk' based on numeric index
\r
973 Chunk& currentChunk = mergedChunks[i];
\r
975 // get iteratorChunk based on vector iterator
\r
976 Chunk& iteratorChunk = (*chunkIter);
\r
978 // if currentChunk.Stop(shifted) == iterator Chunk.Start(shifted)
\r
979 if ( currentChunk.Stop>>16 == iteratorChunk.Start>>16 ) {
\r
981 // set currentChunk.Stop to iteratorChunk.Stop
\r
982 currentChunk.Stop = iteratorChunk.Stop;
\r
987 // set currentChunk + 1 to iteratorChunk
\r
988 mergedChunks.push_back(iteratorChunk);
\r
993 // saved merged chunk vector
\r
994 (*binIter).second = mergedChunks;
\r
999 // opens BAM file (and index)
\r
1000 void BamReader::BamReaderPrivate::Open(const string& filename, const string& indexFilename) {
\r
1002 Filename = filename;
\r
1003 IndexFilename = indexFilename;
\r
1005 // open the BGZF file for reading, retrieve header text & reference data
\r
1006 mBGZF.Open(filename, "rb");
\r
1008 LoadReferenceData();
\r
1010 // store file offset of first alignment
\r
1011 AlignmentsBeginOffset = mBGZF.Tell();
\r
1013 // open index file & load index data (if exists)
\r
1014 if ( !IndexFilename.empty() ) {
\r
1019 // returns BAM file pointer to beginning of alignment data
\r
1020 bool BamReader::BamReaderPrivate::Rewind(void) {
\r
1022 // find first reference that has alignments in the BAM file
\r
1024 int refCount = References.size();
\r
1025 for ( ; refID < refCount; ++refID ) {
\r
1026 if ( References.at(refID).RefHasAlignments ) { break; }
\r
1029 // store default bounds for first alignment
\r
1030 CurrentRefID = refID;
\r
1032 IsRegionSpecified = false;
\r
1034 // return success/failure of seek
\r
1035 return mBGZF.Seek(AlignmentsBeginOffset);
\r
1038 // saves index data to BAM index file (".bai"), returns success/fail
\r
1039 bool BamReader::BamReaderPrivate::WriteIndex(void) {
\r
1041 IndexFilename = Filename + ".bai";
\r
1042 FILE* indexStream = fopen(IndexFilename.c_str(), "wb");
\r
1043 if ( indexStream == 0 ) {
\r
1044 printf("ERROR: Could not open file to save index\n");
\r
1048 // write BAM index header
\r
1049 fwrite("BAI\1", 1, 4, indexStream);
\r
1051 // write number of reference sequences
\r
1052 int32_t numReferenceSeqs = Index.size();
\r
1053 if ( IsBigEndian ) { SwapEndian_32(numReferenceSeqs); }
\r
1054 fwrite(&numReferenceSeqs, 4, 1, indexStream);
\r
1056 // iterate over reference sequences
\r
1057 BamIndex::const_iterator indexIter = Index.begin();
\r
1058 BamIndex::const_iterator indexEnd = Index.end();
\r
1059 for ( ; indexIter != indexEnd; ++ indexIter ) {
\r
1061 // get reference index data
\r
1062 const ReferenceIndex& refIndex = (*indexIter);
\r
1063 const BamBinMap& binMap = refIndex.Bins;
\r
1064 const LinearOffsetVector& offsets = refIndex.Offsets;
\r
1066 // write number of bins
\r
1067 int32_t binCount = binMap.size();
\r
1068 if ( IsBigEndian ) { SwapEndian_32(binCount); }
\r
1069 fwrite(&binCount, 4, 1, indexStream);
\r
1071 // iterate over bins
\r
1072 BamBinMap::const_iterator binIter = binMap.begin();
\r
1073 BamBinMap::const_iterator binEnd = binMap.end();
\r
1074 for ( ; binIter != binEnd; ++binIter ) {
\r
1076 // get bin data (key and chunk vector)
\r
1077 uint32_t binKey = (*binIter).first;
\r
1078 const ChunkVector& binChunks = (*binIter).second;
\r
1080 // save BAM bin key
\r
1081 if ( IsBigEndian ) { SwapEndian_32(binKey); }
\r
1082 fwrite(&binKey, 4, 1, indexStream);
\r
1084 // save chunk count
\r
1085 int32_t chunkCount = binChunks.size();
\r
1086 if ( IsBigEndian ) { SwapEndian_32(chunkCount); }
\r
1087 fwrite(&chunkCount, 4, 1, indexStream);
\r
1089 // iterate over chunks
\r
1090 ChunkVector::const_iterator chunkIter = binChunks.begin();
\r
1091 ChunkVector::const_iterator chunkEnd = binChunks.end();
\r
1092 for ( ; chunkIter != chunkEnd; ++chunkIter ) {
\r
1094 // get current chunk data
\r
1095 const Chunk& chunk = (*chunkIter);
\r
1096 uint64_t start = chunk.Start;
\r
1097 uint64_t stop = chunk.Stop;
\r
1099 if ( IsBigEndian ) {
\r
1100 SwapEndian_64(start);
\r
1101 SwapEndian_64(stop);
\r
1104 // save chunk offsets
\r
1105 fwrite(&start, 8, 1, indexStream);
\r
1106 fwrite(&stop, 8, 1, indexStream);
\r
1110 // write linear offsets size
\r
1111 int32_t offsetSize = offsets.size();
\r
1112 if ( IsBigEndian ) { SwapEndian_32(offsetSize); }
\r
1113 fwrite(&offsetSize, 4, 1, indexStream);
\r
1115 // iterate over linear offsets
\r
1116 LinearOffsetVector::const_iterator offsetIter = offsets.begin();
\r
1117 LinearOffsetVector::const_iterator offsetEnd = offsets.end();
\r
1118 for ( ; offsetIter != offsetEnd; ++offsetIter ) {
\r
1120 // write linear offset value
\r
1121 uint64_t linearOffset = (*offsetIter);
\r
1122 if ( IsBigEndian ) { SwapEndian_64(linearOffset); }
\r
1123 fwrite(&linearOffset, 8, 1, indexStream);
\r
1127 // flush buffer, close file, and return success
\r
1128 fflush(indexStream);
\r
1129 fclose(indexStream);
\r