1 // ***************************************************************************
\r
2 // BamReader.cpp (c) 2009 Derek Barnett, Michael Strömberg
\r
3 // Marth Lab, Department of Biology, Boston College
\r
4 // All rights reserved.
\r
5 // ---------------------------------------------------------------------------
\r
6 // Last modified: 11 January 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 struct BamReader::BamReaderPrivate {
\r
28 // -------------------------------
\r
30 // -------------------------------
\r
36 RefVector References;
\r
38 int64_t AlignmentsBeginOffset;
\r
40 string IndexFilename;
\r
42 // user-specified region values
\r
43 bool IsRegionSpecified;
\r
47 // BAM character constants
\r
48 const char* DNA_LOOKUP;
\r
49 const char* CIGAR_LOOKUP;
\r
51 // -------------------------------
\r
52 // constructor & destructor
\r
53 // -------------------------------
\r
54 BamReaderPrivate(void);
\r
55 ~BamReaderPrivate(void);
\r
57 // -------------------------------
\r
58 // "public" interface
\r
59 // -------------------------------
\r
63 bool Jump(int refID, int position = 0);
\r
64 void Open(const string& filename, const string& indexFilename = "");
\r
67 // access alignment data
\r
68 bool GetNextAlignment(BamAlignment& bAlignment);
\r
70 // access auxiliary data
\r
71 const string GetHeaderText(void) const;
\r
72 const int GetReferenceCount(void) const;
\r
73 const RefVector GetReferenceData(void) const;
\r
74 const int GetReferenceID(const string& refName) const;
\r
77 bool CreateIndex(void);
\r
79 // -------------------------------
\r
81 // -------------------------------
\r
83 // *** reading alignments and auxiliary data *** //
\r
85 // calculate bins that overlap region ( left to reference end for now )
\r
86 int BinsFromRegion(int refID, int left, uint16_t[MAX_BIN]);
\r
87 // calculates alignment end position based on starting position and provided CIGAR operations
\r
88 int CalculateAlignmentEnd(const int& position, const std::vector<CigarOp>& cigarData);
\r
89 // calculate file offset for first alignment chunk overlapping 'left'
\r
90 int64_t GetOffset(int refID, int left);
\r
91 // checks to see if alignment overlaps current region
\r
92 bool IsOverlap(BamAlignment& bAlignment);
\r
93 // retrieves header text from BAM file
\r
94 void LoadHeaderData(void);
\r
95 // retrieves BAM alignment under file pointer
\r
96 bool LoadNextAlignment(BamAlignment& bAlignment);
\r
97 // builds reference data structure from BAM file
\r
98 void LoadReferenceData(void);
\r
100 // *** index file handling *** //
\r
102 // calculates index for BAM file
\r
103 bool BuildIndex(void);
\r
104 // clear out inernal index data structure
\r
105 void ClearIndex(void);
\r
106 // saves BAM bin entry for index
\r
107 void InsertBinEntry(BamBinMap& binMap, const uint32_t& saveBin, const uint64_t& saveOffset, const uint64_t& lastOffset);
\r
108 // saves linear offset entry for index
\r
109 void InsertLinearOffset(LinearOffsetVector& offsets, const BamAlignment& bAlignment, const uint64_t& lastOffset);
\r
110 // loads index from BAM index file
\r
111 bool LoadIndex(void);
\r
112 // simplifies index by merging 'chunks'
\r
113 void MergeChunks(void);
\r
114 // round-up 32-bit integer to next power-of-2
\r
115 void Roundup32(int& value);
\r
116 // saves index to BAM index file
\r
117 bool WriteIndex(void);
\r
120 // -----------------------------------------------------
\r
121 // BamReader implementation (wrapper around BRPrivate)
\r
122 // -----------------------------------------------------
\r
125 BamReader::BamReader(void) {
\r
126 d = new BamReaderPrivate;
\r
130 BamReader::~BamReader(void) {
\r
136 void BamReader::Close(void) { d->Close(); }
\r
137 bool BamReader::Jump(int refID, int position) { return d->Jump(refID, position); }
\r
138 void BamReader::Open(const string& filename, const string& indexFilename) { d->Open(filename, indexFilename); }
\r
139 bool BamReader::Rewind(void) { return d->Rewind(); }
\r
141 // access alignment data
\r
142 bool BamReader::GetNextAlignment(BamAlignment& bAlignment) { return d->GetNextAlignment(bAlignment); }
\r
144 // access auxiliary data
\r
145 const string BamReader::GetHeaderText(void) const { return d->HeaderText; }
\r
146 const int BamReader::GetReferenceCount(void) const { return d->References.size(); }
\r
147 const RefVector BamReader::GetReferenceData(void) const { return d->References; }
\r
148 const int BamReader::GetReferenceID(const string& refName) const { return d->GetReferenceID(refName); }
\r
150 // index operations
\r
151 bool BamReader::CreateIndex(void) { return d->CreateIndex(); }
\r
153 // -----------------------------------------------------
\r
154 // BamReaderPrivate implementation
\r
155 // -----------------------------------------------------
\r
158 BamReader::BamReaderPrivate::BamReaderPrivate(void)
\r
159 : IsIndexLoaded(false)
\r
160 , AlignmentsBeginOffset(0)
\r
161 , IsRegionSpecified(false)
\r
164 , DNA_LOOKUP("=ACMGRSVTWYHKDBN")
\r
165 , CIGAR_LOOKUP("MIDNSHP")
\r
169 BamReader::BamReaderPrivate::~BamReaderPrivate(void) {
\r
173 // calculate bins that overlap region ( left to reference end for now )
\r
174 int BamReader::BamReaderPrivate::BinsFromRegion(int refID, int left, uint16_t list[MAX_BIN]) {
\r
176 // get region boundaries
\r
177 uint32_t begin = (unsigned int)left;
\r
178 uint32_t end = (unsigned int)References.at(refID).RefLength - 1;
\r
180 // initialize list, bin '0' always a valid bin
\r
184 // get rest of bins that contain this region
\r
186 for (k = 1 + (begin>>26); k <= 1 + (end>>26); ++k) { list[i++] = k; }
\r
187 for (k = 9 + (begin>>23); k <= 9 + (end>>23); ++k) { list[i++] = k; }
\r
188 for (k = 73 + (begin>>20); k <= 73 + (end>>20); ++k) { list[i++] = k; }
\r
189 for (k = 585 + (begin>>17); k <= 585 + (end>>17); ++k) { list[i++] = k; }
\r
190 for (k = 4681 + (begin>>14); k <= 4681 + (end>>14); ++k) { list[i++] = k; }
\r
192 // return number of bins stored
\r
196 // populates BAM index data structure from BAM file data
\r
197 bool BamReader::BamReaderPrivate::BuildIndex(void) {
\r
199 // check to be sure file is open
\r
200 if (!mBGZF.IsOpen) { return false; }
\r
202 // move file pointer to beginning of alignments
\r
205 // get reference count, reserve index space
\r
206 int numReferences = References.size();
\r
207 for ( int i = 0; i < numReferences; ++i ) {
\r
208 Index.push_back(ReferenceIndex());
\r
211 // sets default constant for bin, ID, offset, coordinate variables
\r
212 const uint32_t defaultValue = 0xffffffffu;
\r
215 uint32_t saveBin(defaultValue);
\r
216 uint32_t lastBin(defaultValue);
\r
218 // reference ID data
\r
219 int32_t saveRefID(defaultValue);
\r
220 int32_t lastRefID(defaultValue);
\r
223 uint64_t saveOffset = mBGZF.Tell();
\r
224 uint64_t lastOffset = saveOffset;
\r
227 int32_t lastCoordinate = defaultValue;
\r
229 BamAlignment bAlignment;
\r
230 while( GetNextAlignment(bAlignment) ) {
\r
232 // change of chromosome, save ID, reset bin
\r
233 if ( lastRefID != bAlignment.RefID ) {
\r
234 lastRefID = bAlignment.RefID;
\r
235 lastBin = defaultValue;
\r
238 // if lastCoordinate greater than BAM position - file not sorted properly
\r
239 else if ( lastCoordinate > bAlignment.Position ) {
\r
240 printf("BAM file not properly sorted:\n");
\r
241 printf("Alignment %s : %d > %d on reference (id = %d)", bAlignment.Name.c_str(), lastCoordinate, bAlignment.Position, bAlignment.RefID);
\r
245 // if valid reference && BAM bin spans some minimum cutoff (smaller bin ids span larger regions)
\r
246 if ( (bAlignment.RefID >= 0) && (bAlignment.Bin < 4681) ) {
\r
248 // save linear offset entry (matched to BAM entry refID)
\r
249 ReferenceIndex& refIndex = Index.at(bAlignment.RefID);
\r
250 LinearOffsetVector& offsets = refIndex.Offsets;
\r
251 InsertLinearOffset(offsets, bAlignment, lastOffset);
\r
254 // if current BamAlignment bin != lastBin, "then possibly write the binning index"
\r
255 if ( bAlignment.Bin != lastBin ) {
\r
257 // if not first time through
\r
258 if ( saveBin != defaultValue ) {
\r
260 // save Bam bin entry
\r
261 ReferenceIndex& refIndex = Index.at(saveRefID);
\r
262 BamBinMap& binMap = refIndex.Bins;
\r
263 InsertBinEntry(binMap, saveBin, saveOffset, lastOffset);
\r
266 // update saveOffset
\r
267 saveOffset = lastOffset;
\r
269 // update bin values
\r
270 saveBin = bAlignment.Bin;
\r
271 lastBin = bAlignment.Bin;
\r
273 // update saveRefID
\r
274 saveRefID = bAlignment.RefID;
\r
276 // if invalid RefID, break out (why?)
\r
277 if ( saveRefID < 0 ) { break; }
\r
280 // make sure that current file pointer is beyond lastOffset
\r
281 if ( mBGZF.Tell() <= (int64_t)lastOffset ) {
\r
282 printf("Error in BGZF offsets.\n");
\r
286 // update lastOffset
\r
287 lastOffset = mBGZF.Tell();
\r
289 // update lastCoordinate
\r
290 lastCoordinate = bAlignment.Position;
\r
293 // save any leftover BAM data (as long as refID is valid)
\r
294 if ( saveRefID >= 0 ) {
\r
295 // save Bam bin entry
\r
296 ReferenceIndex& refIndex = Index.at(saveRefID);
\r
297 BamBinMap& binMap = refIndex.Bins;
\r
298 InsertBinEntry(binMap, saveBin, saveOffset, lastOffset);
\r
301 // simplify index by merging chunks
\r
304 // iterate over references
\r
305 BamIndex::iterator indexIter = Index.begin();
\r
306 BamIndex::iterator indexEnd = Index.end();
\r
307 for ( int i = 0; indexIter != indexEnd; ++indexIter, ++i ) {
\r
309 // get reference index data
\r
310 ReferenceIndex& refIndex = (*indexIter);
\r
311 BamBinMap& binMap = refIndex.Bins;
\r
312 LinearOffsetVector& offsets = refIndex.Offsets;
\r
314 // store whether reference has alignments or no
\r
315 References[i].RefHasAlignments = ( binMap.size() > 0 );
\r
317 // sort linear offsets
\r
318 sort(offsets.begin(), offsets.end());
\r
322 // rewind file pointer to beginning of alignments, return success/fail
\r
326 // calculates alignment end position based on starting position and provided CIGAR operations
\r
327 int BamReader::BamReaderPrivate::CalculateAlignmentEnd(const int& position, const vector<CigarOp>& cigarData) {
\r
329 // initialize alignment end to starting position
\r
330 int alignEnd = position;
\r
332 // iterate over cigar operations
\r
333 vector<CigarOp>::const_iterator cigarIter = cigarData.begin();
\r
334 vector<CigarOp>::const_iterator cigarEnd = cigarData.end();
\r
335 for ( ; cigarIter != cigarEnd; ++cigarIter) {
\r
336 char cigarType = (*cigarIter).Type;
\r
337 if ( cigarType == 'M' || cigarType == 'D' || cigarType == 'N' ) {
\r
338 alignEnd += (*cigarIter).Length;
\r
345 // clear index data structure
\r
346 void BamReader::BamReaderPrivate::ClearIndex(void) {
\r
347 Index.clear(); // sufficient ??
\r
350 // closes the BAM file
\r
351 void BamReader::BamReaderPrivate::Close(void) {
\r
354 HeaderText.clear();
\r
355 IsRegionSpecified = false;
\r
358 // create BAM index from BAM file (keep structure in memory) and write to default index output file
\r
359 bool BamReader::BamReaderPrivate::CreateIndex(void) {
\r
364 // build (& save) index from BAM file
\r
366 ok &= BuildIndex();
\r
367 ok &= WriteIndex();
\r
369 // return success/fail
\r
373 // returns RefID for given RefName (returns References.size() if not found)
\r
374 const int BamReader::BamReaderPrivate::GetReferenceID(const string& refName) const {
\r
376 // retrieve names from reference data
\r
377 vector<string> refNames;
\r
378 RefVector::const_iterator refIter = References.begin();
\r
379 RefVector::const_iterator refEnd = References.end();
\r
380 for ( ; refIter != refEnd; ++refIter) {
\r
381 refNames.push_back( (*refIter).RefName );
\r
384 // return 'index-of' refName ( if not found, returns refNames.size() )
\r
385 return distance(refNames.begin(), find(refNames.begin(), refNames.end(), refName));
\r
388 // get next alignment (from specified region, if given)
\r
389 bool BamReader::BamReaderPrivate::GetNextAlignment(BamAlignment& bAlignment) {
\r
391 // if valid alignment available
\r
392 if ( LoadNextAlignment(bAlignment) ) {
\r
394 // if region not specified, return success
\r
395 if ( !IsRegionSpecified ) { return true; }
\r
397 // load next alignment until region overlap is found
\r
398 while ( !IsOverlap(bAlignment) ) {
\r
399 // if no valid alignment available (likely EOF) return failure
\r
400 if ( !LoadNextAlignment(bAlignment) ) { return false; }
\r
403 // return success (alignment found that overlaps region)
\r
407 // no valid alignment
\r
408 else { return false; }
\r
411 // calculate closest indexed file offset for region specified
\r
412 int64_t BamReader::BamReaderPrivate::GetOffset(int refID, int left) {
\r
414 // calculate which bins overlap this region
\r
415 uint16_t* bins = (uint16_t*)calloc(MAX_BIN, 2);
\r
416 int numBins = BinsFromRegion(refID, left, bins);
\r
418 // get bins for this reference
\r
419 const ReferenceIndex& refIndex = Index.at(refID);
\r
420 const BamBinMap& binMap = refIndex.Bins;
\r
422 // get minimum offset to consider
\r
423 const LinearOffsetVector& offsets = refIndex.Offsets;
\r
424 uint64_t minOffset = ( (unsigned int)(left>>BAM_LIDX_SHIFT) >= offsets.size() ) ? 0 : offsets.at(left>>BAM_LIDX_SHIFT);
\r
426 // store offsets to beginning of alignment 'chunks'
\r
427 std::vector<int64_t> chunkStarts;
\r
429 // store all alignment 'chunk' starts for bins in this region
\r
430 for (int i = 0; i < numBins; ++i ) {
\r
431 uint16_t binKey = bins[i];
\r
433 map<uint32_t, ChunkVector>::const_iterator binIter = binMap.find(binKey);
\r
434 if ( (binIter != binMap.end()) && ((*binIter).first == binKey) ) {
\r
436 const ChunkVector& chunks = (*binIter).second;
\r
437 std::vector<Chunk>::const_iterator chunksIter = chunks.begin();
\r
438 std::vector<Chunk>::const_iterator chunksEnd = chunks.end();
\r
439 for ( ; chunksIter != chunksEnd; ++chunksIter) {
\r
440 const Chunk& chunk = (*chunksIter);
\r
441 if ( chunk.Stop > minOffset ) {
\r
442 chunkStarts.push_back( chunk.Start );
\r
451 // if no alignments found, else return smallest offset for alignment starts
\r
452 if ( chunkStarts.size() == 0 ) { return -1; }
\r
453 else { return *min_element(chunkStarts.begin(), chunkStarts.end()); }
\r
456 // saves BAM bin entry for index
\r
457 void BamReader::BamReaderPrivate::InsertBinEntry(BamBinMap& binMap,
\r
458 const uint32_t& saveBin,
\r
459 const uint64_t& saveOffset,
\r
460 const uint64_t& lastOffset)
\r
463 BamBinMap::iterator binIter = binMap.find(saveBin);
\r
465 // create new chunk
\r
466 Chunk newChunk(saveOffset, lastOffset);
\r
468 // if entry doesn't exist
\r
469 if ( binIter == binMap.end() ) {
\r
470 ChunkVector newChunks;
\r
471 newChunks.push_back(newChunk);
\r
472 binMap.insert( pair<uint32_t, ChunkVector>(saveBin, newChunks));
\r
477 ChunkVector& binChunks = (*binIter).second;
\r
478 binChunks.push_back( newChunk );
\r
482 // saves linear offset entry for index
\r
483 void BamReader::BamReaderPrivate::InsertLinearOffset(LinearOffsetVector& offsets,
\r
484 const BamAlignment& bAlignment,
\r
485 const uint64_t& lastOffset)
\r
487 // get converted offsets
\r
488 int beginOffset = bAlignment.Position >> BAM_LIDX_SHIFT;
\r
489 int endOffset = ( CalculateAlignmentEnd(bAlignment.Position, bAlignment.CigarData) - 1) >> BAM_LIDX_SHIFT;
\r
491 // resize vector if necessary
\r
492 int oldSize = offsets.size();
\r
493 int newSize = endOffset + 1;
\r
494 if ( oldSize < newSize ) {
\r
495 Roundup32(newSize);
\r
496 offsets.resize(newSize, 0);
\r
500 for(int i = beginOffset + 1; i <= endOffset ; ++i) {
\r
501 if ( offsets[i] == 0) {
\r
502 offsets[i] = lastOffset;
\r
507 // returns whether alignment overlaps currently specified region (refID, leftBound)
\r
508 bool BamReader::BamReaderPrivate::IsOverlap(BamAlignment& bAlignment) {
\r
510 // if on different reference sequence, quit
\r
511 if ( bAlignment.RefID != CurrentRefID ) { return false; }
\r
513 // read starts after left boundary
\r
514 if ( bAlignment.Position >= CurrentLeft) { return true; }
\r
516 // return whether alignment end overlaps left boundary
\r
517 return ( CalculateAlignmentEnd(bAlignment.Position, bAlignment.CigarData) >= CurrentLeft );
\r
520 // jumps to specified region(refID, leftBound) in BAM file, returns success/fail
\r
521 bool BamReader::BamReaderPrivate::Jump(int refID, int position) {
\r
523 // if data exists for this reference and position is valid
\r
524 if ( References.at(refID).RefHasAlignments && (position <= References.at(refID).RefLength) ) {
\r
526 // set current region
\r
527 CurrentRefID = refID;
\r
528 CurrentLeft = position;
\r
529 IsRegionSpecified = true;
\r
531 // calculate offset
\r
532 int64_t offset = GetOffset(CurrentRefID, CurrentLeft);
\r
534 // if in valid offset, return failure
\r
535 if ( offset == -1 ) { return false; }
\r
537 // otherwise return success of seek operation
\r
538 else { return mBGZF.Seek(offset); }
\r
541 // invalid jump request parameters, return failure
\r
545 // load BAM header data
\r
546 void BamReader::BamReaderPrivate::LoadHeaderData(void) {
\r
548 // check to see if proper BAM header
\r
550 if (mBGZF.Read(buffer, 4) != 4) {
\r
551 printf("Could not read header type\n");
\r
555 if (strncmp(buffer, "BAM\001", 4)) {
\r
556 printf("wrong header type!\n");
\r
560 // get BAM header text length
\r
561 mBGZF.Read(buffer, 4);
\r
562 const unsigned int headerTextLength = BgzfData::UnpackUnsignedInt(buffer);
\r
564 // get BAM header text
\r
565 char* headerText = (char*)calloc(headerTextLength + 1, 1);
\r
566 mBGZF.Read(headerText, headerTextLength);
\r
567 HeaderText = (string)((const char*)headerText);
\r
569 // clean up calloc-ed temp variable
\r
573 // load existing index data from BAM index file (".bai"), return success/fail
\r
574 bool BamReader::BamReaderPrivate::LoadIndex(void) {
\r
576 // clear out index data
\r
579 // skip if index file empty
\r
580 if ( IndexFilename.empty() ) { return false; }
\r
582 // open index file, abort on error
\r
583 FILE* indexStream = fopen(IndexFilename.c_str(), "rb");
\r
585 printf("ERROR: Unable to open the BAM index file %s for reading.\n", IndexFilename.c_str() );
\r
589 size_t elementsRead = 0;
\r
591 // see if index is valid BAM index
\r
593 elementsRead = fread(magic, 1, 4, indexStream);
\r
594 if (strncmp(magic, "BAI\1", 4)) {
\r
595 printf("Problem with index file - invalid format.\n");
\r
596 fclose(indexStream);
\r
600 // get number of reference sequences
\r
601 uint32_t numRefSeqs;
\r
602 elementsRead = fread(&numRefSeqs, 4, 1, indexStream);
\r
604 // intialize space for BamIndex data structure
\r
605 Index.reserve(numRefSeqs);
\r
607 // iterate over reference sequences
\r
608 for (unsigned int i = 0; i < numRefSeqs; ++i) {
\r
610 // get number of bins for this reference sequence
\r
612 elementsRead = fread(&numBins, 4, 1, indexStream);
\r
615 RefData& refEntry = References[i];
\r
616 refEntry.RefHasAlignments = true;
\r
619 // intialize BinVector
\r
622 // iterate over bins for that reference sequence
\r
623 for (int j = 0; j < numBins; ++j) {
\r
627 elementsRead = fread(&binID, 4, 1, indexStream);
\r
629 // get number of regionChunks in this bin
\r
630 uint32_t numChunks;
\r
631 elementsRead = fread(&numChunks, 4, 1, indexStream);
\r
633 // intialize ChunkVector
\r
634 ChunkVector regionChunks;
\r
635 regionChunks.reserve(numChunks);
\r
637 // iterate over regionChunks in this bin
\r
638 for (unsigned int k = 0; k < numChunks; ++k) {
\r
640 // get chunk boundaries (left, right)
\r
643 elementsRead = fread(&left, 8, 1, indexStream);
\r
644 elementsRead = fread(&right, 8, 1, indexStream);
\r
647 regionChunks.push_back( Chunk(left, right) );
\r
650 // sort chunks for this bin
\r
651 sort( regionChunks.begin(), regionChunks.end(), ChunkLessThan );
\r
653 // save binID, chunkVector for this bin
\r
654 binMap.insert( pair<uint32_t, ChunkVector>(binID, regionChunks) );
\r
657 // load linear index for this reference sequence
\r
659 // get number of linear offsets
\r
660 int32_t numLinearOffsets;
\r
661 elementsRead = fread(&numLinearOffsets, 4, 1, indexStream);
\r
663 // intialize LinearOffsetVector
\r
664 LinearOffsetVector offsets;
\r
665 offsets.reserve(numLinearOffsets);
\r
667 // iterate over linear offsets for this reference sequeence
\r
668 uint64_t linearOffset;
\r
669 for (int j = 0; j < numLinearOffsets; ++j) {
\r
670 // read a linear offset & store
\r
671 elementsRead = fread(&linearOffset, 8, 1, indexStream);
\r
672 offsets.push_back(linearOffset);
\r
675 // sort linear offsets
\r
676 sort( offsets.begin(), offsets.end() );
\r
678 // store index data for that reference sequence
\r
679 Index.push_back( ReferenceIndex(binMap, offsets) );
\r
682 // close index file (.bai) and return
\r
683 fclose(indexStream);
\r
687 // populates BamAlignment with alignment data under file pointer, returns success/fail
\r
688 bool BamReader::BamReaderPrivate::LoadNextAlignment(BamAlignment& bAlignment) {
\r
690 // read in the 'block length' value, make sure it's not zero
\r
692 mBGZF.Read(buffer, 4);
\r
693 const unsigned int blockLength = BgzfData::UnpackUnsignedInt(buffer);
\r
694 if ( blockLength == 0 ) { return false; }
\r
696 // keep track of bytes read as method progresses
\r
699 // read in core alignment data, make sure the right size of data was read
\r
700 char x[BAM_CORE_SIZE];
\r
701 if ( mBGZF.Read(x, BAM_CORE_SIZE) != BAM_CORE_SIZE ) { return false; }
\r
702 bytesRead += BAM_CORE_SIZE;
\r
704 // set BamAlignment 'core' data and character data lengths
\r
705 unsigned int tempValue;
\r
706 unsigned int queryNameLength;
\r
707 unsigned int numCigarOperations;
\r
708 unsigned int querySequenceLength;
\r
710 bAlignment.RefID = BgzfData::UnpackSignedInt(&x[0]);
\r
711 bAlignment.Position = BgzfData::UnpackSignedInt(&x[4]);
\r
713 tempValue = BgzfData::UnpackUnsignedInt(&x[8]);
\r
714 bAlignment.Bin = tempValue >> 16;
\r
715 bAlignment.MapQuality = tempValue >> 8 & 0xff;
\r
716 queryNameLength = tempValue & 0xff;
\r
718 tempValue = BgzfData::UnpackUnsignedInt(&x[12]);
\r
719 bAlignment.AlignmentFlag = tempValue >> 16;
\r
720 numCigarOperations = tempValue & 0xffff;
\r
722 querySequenceLength = BgzfData::UnpackUnsignedInt(&x[16]);
\r
723 bAlignment.MateRefID = BgzfData::UnpackSignedInt(&x[20]);
\r
724 bAlignment.MatePosition = BgzfData::UnpackSignedInt(&x[24]);
\r
725 bAlignment.InsertSize = BgzfData::UnpackSignedInt(&x[28]);
\r
727 // calculate lengths/offsets
\r
728 const unsigned int dataLength = blockLength - BAM_CORE_SIZE;
\r
729 const unsigned int cigarDataOffset = queryNameLength;
\r
730 const unsigned int seqDataOffset = cigarDataOffset + (numCigarOperations * 4);
\r
731 const unsigned int qualDataOffset = seqDataOffset + (querySequenceLength+1)/2;
\r
732 const unsigned int tagDataOffset = qualDataOffset + querySequenceLength;
\r
733 const unsigned int tagDataLen = dataLength - tagDataOffset;
\r
735 // set up destination buffers for character data
\r
736 char* allCharData = (char*)calloc(sizeof(char), dataLength);
\r
737 uint32_t* cigarData = (uint32_t*)(allCharData + cigarDataOffset);
\r
738 char* seqData = ((char*)allCharData) + seqDataOffset;
\r
739 char* qualData = ((char*)allCharData) + qualDataOffset;
\r
740 char* tagData = ((char*)allCharData) + tagDataOffset;
\r
742 // get character data - make sure proper data size was read
\r
743 if ( mBGZF.Read(allCharData, dataLength) != (signed int)dataLength) { return false; }
\r
746 bytesRead += dataLength;
\r
748 // clear out any previous string data
\r
749 bAlignment.Name.clear();
\r
750 bAlignment.QueryBases.clear();
\r
751 bAlignment.Qualities.clear();
\r
752 bAlignment.AlignedBases.clear();
\r
753 bAlignment.CigarData.clear();
\r
754 bAlignment.TagData.clear();
\r
757 bAlignment.Name = (string)((const char*)(allCharData));
\r
759 // save query sequence
\r
760 for (unsigned int i = 0; i < querySequenceLength; ++i) {
\r
761 char singleBase = DNA_LOOKUP[ ( ( seqData[(i/2)] >> (4*(1-(i%2)))) & 0xf ) ];
\r
762 bAlignment.QueryBases.append( 1, singleBase );
\r
765 // save sequence length
\r
766 bAlignment.Length = bAlignment.QueryBases.length();
\r
768 // save qualities, convert from numeric QV to FASTQ character
\r
769 for (unsigned int i = 0; i < querySequenceLength; ++i) {
\r
770 char singleQuality = (char)(qualData[i]+33);
\r
771 bAlignment.Qualities.append( 1, singleQuality );
\r
774 // save CIGAR-related data;
\r
776 for (unsigned int i = 0; i < numCigarOperations; ++i) {
\r
778 // build CigarOp struct
\r
780 op.Length = (cigarData[i] >> BAM_CIGAR_SHIFT);
\r
781 op.Type = CIGAR_LOOKUP[ (cigarData[i] & BAM_CIGAR_MASK) ];
\r
784 bAlignment.CigarData.push_back(op);
\r
786 // build AlignedBases string
\r
790 case ('I') : bAlignment.AlignedBases.append( bAlignment.QueryBases.substr(k, op.Length) ); // for 'M', 'I' - write bases
\r
791 case ('S') : k += op.Length; // for 'S' - skip over query bases
\r
794 case ('D') : bAlignment.AlignedBases.append( op.Length, '-' ); // for 'D' - write gap character
\r
797 case ('P') : bAlignment.AlignedBases.append( op.Length, '*' ); // for 'P' - write padding character;
\r
800 case ('N') : bAlignment.AlignedBases.append( op.Length, 'N' ); // for 'N' - write N's, skip bases in query sequence
\r
804 case ('H') : break; // for 'H' - do nothing, move to next op
\r
806 default : printf("ERROR: Invalid Cigar op type\n"); // shouldn't get here
\r
811 // read in the tag data
\r
812 bAlignment.TagData.resize(tagDataLen);
\r
813 memcpy((char*)bAlignment.TagData.data(), tagData, tagDataLen);
\r
820 // loads reference data from BAM file
\r
821 void BamReader::BamReaderPrivate::LoadReferenceData(void) {
\r
823 // get number of reference sequences
\r
825 mBGZF.Read(buffer, 4);
\r
826 const unsigned int numberRefSeqs = BgzfData::UnpackUnsignedInt(buffer);
\r
827 if (numberRefSeqs == 0) { return; }
\r
828 References.reserve((int)numberRefSeqs);
\r
830 // iterate over all references in header
\r
831 for (unsigned int i = 0; i != numberRefSeqs; ++i) {
\r
833 // get length of reference name
\r
834 mBGZF.Read(buffer, 4);
\r
835 const unsigned int refNameLength = BgzfData::UnpackUnsignedInt(buffer);
\r
836 char* refName = (char*)calloc(refNameLength, 1);
\r
838 // get reference name and reference sequence length
\r
839 mBGZF.Read(refName, refNameLength);
\r
840 mBGZF.Read(buffer, 4);
\r
841 const int refLength = BgzfData::UnpackSignedInt(buffer);
\r
843 // store data for reference
\r
844 RefData aReference;
\r
845 aReference.RefName = (string)((const char*)refName);
\r
846 aReference.RefLength = refLength;
\r
847 References.push_back(aReference);
\r
849 // clean up calloc-ed temp variable
\r
854 // merges 'alignment chunks' in BAM bin (used for index building)
\r
855 void BamReader::BamReaderPrivate::MergeChunks(void) {
\r
857 // iterate over reference enties
\r
858 BamIndex::iterator indexIter = Index.begin();
\r
859 BamIndex::iterator indexEnd = Index.end();
\r
860 for ( ; indexIter != indexEnd; ++indexIter ) {
\r
862 // get BAM bin map for this reference
\r
863 ReferenceIndex& refIndex = (*indexIter);
\r
864 BamBinMap& bamBinMap = refIndex.Bins;
\r
866 // iterate over BAM bins
\r
867 BamBinMap::iterator binIter = bamBinMap.begin();
\r
868 BamBinMap::iterator binEnd = bamBinMap.end();
\r
869 for ( ; binIter != binEnd; ++binIter ) {
\r
871 // get chunk vector for this bin
\r
872 ChunkVector& binChunks = (*binIter).second;
\r
873 if ( binChunks.size() == 0 ) { continue; }
\r
875 ChunkVector mergedChunks;
\r
876 mergedChunks.push_back( binChunks[0] );
\r
878 // iterate over chunks
\r
880 ChunkVector::iterator chunkIter = binChunks.begin();
\r
881 ChunkVector::iterator chunkEnd = binChunks.end();
\r
882 for ( ++chunkIter; chunkIter != chunkEnd; ++chunkIter) {
\r
884 // get 'currentChunk' based on numeric index
\r
885 Chunk& currentChunk = mergedChunks[i];
\r
887 // get iteratorChunk based on vector iterator
\r
888 Chunk& iteratorChunk = (*chunkIter);
\r
890 // if currentChunk.Stop(shifted) == iterator Chunk.Start(shifted)
\r
891 if ( currentChunk.Stop>>16 == iteratorChunk.Start>>16 ) {
\r
893 // set currentChunk.Stop to iteratorChunk.Stop
\r
894 currentChunk.Stop = iteratorChunk.Stop;
\r
899 // set currentChunk + 1 to iteratorChunk
\r
900 mergedChunks.push_back(iteratorChunk);
\r
905 // saved merged chunk vector
\r
906 (*binIter).second = mergedChunks;
\r
911 // opens BAM file (and index)
\r
912 void BamReader::BamReaderPrivate::Open(const string& filename, const string& indexFilename) {
\r
914 Filename = filename;
\r
915 IndexFilename = indexFilename;
\r
917 // open the BGZF file for reading, retrieve header text & reference data
\r
918 mBGZF.Open(filename, "rb");
\r
920 LoadReferenceData();
\r
922 // store file offset of first alignment
\r
923 AlignmentsBeginOffset = mBGZF.Tell();
\r
925 // open index file & load index data (if exists)
\r
926 if ( !IndexFilename.empty() ) {
\r
931 // returns BAM file pointer to beginning of alignment data
\r
932 bool BamReader::BamReaderPrivate::Rewind(void) {
\r
934 // find first reference that has alignments in the BAM file
\r
936 int refCount = References.size();
\r
937 for ( ; refID < refCount; ++refID ) {
\r
938 if ( References.at(refID).RefHasAlignments ) { break; }
\r
941 // store default bounds for first alignment
\r
942 CurrentRefID = refID;
\r
944 IsRegionSpecified = false;
\r
946 // return success/failure of seek
\r
947 return mBGZF.Seek(AlignmentsBeginOffset);
\r
950 // rounds value up to next power-of-2 (used in index building)
\r
951 void BamReader::BamReaderPrivate::Roundup32(int& value) {
\r
953 value |= value >> 1;
\r
954 value |= value >> 2;
\r
955 value |= value >> 4;
\r
956 value |= value >> 8;
\r
957 value |= value >> 16;
\r
961 // saves index data to BAM index file (".bai"), returns success/fail
\r
962 bool BamReader::BamReaderPrivate::WriteIndex(void) {
\r
964 IndexFilename = Filename + ".bai";
\r
965 FILE* indexStream = fopen(IndexFilename.c_str(), "wb");
\r
966 if ( indexStream == 0 ) {
\r
967 printf("ERROR: Could not open file to save index\n");
\r
971 // write BAM index header
\r
972 fwrite("BAI\1", 1, 4, indexStream);
\r
974 // write number of reference sequences
\r
975 int32_t numReferenceSeqs = Index.size();
\r
976 fwrite(&numReferenceSeqs, 4, 1, indexStream);
\r
978 // iterate over reference sequences
\r
979 BamIndex::const_iterator indexIter = Index.begin();
\r
980 BamIndex::const_iterator indexEnd = Index.end();
\r
981 for ( ; indexIter != indexEnd; ++ indexIter ) {
\r
983 // get reference index data
\r
984 const ReferenceIndex& refIndex = (*indexIter);
\r
985 const BamBinMap& binMap = refIndex.Bins;
\r
986 const LinearOffsetVector& offsets = refIndex.Offsets;
\r
988 // write number of bins
\r
989 int32_t binCount = binMap.size();
\r
990 fwrite(&binCount, 4, 1, indexStream);
\r
992 // iterate over bins
\r
993 BamBinMap::const_iterator binIter = binMap.begin();
\r
994 BamBinMap::const_iterator binEnd = binMap.end();
\r
995 for ( ; binIter != binEnd; ++binIter ) {
\r
997 // get bin data (key and chunk vector)
\r
998 const uint32_t& binKey = (*binIter).first;
\r
999 const ChunkVector& binChunks = (*binIter).second;
\r
1001 // save BAM bin key
\r
1002 fwrite(&binKey, 4, 1, indexStream);
\r
1004 // save chunk count
\r
1005 int32_t chunkCount = binChunks.size();
\r
1006 fwrite(&chunkCount, 4, 1, indexStream);
\r
1008 // iterate over chunks
\r
1009 ChunkVector::const_iterator chunkIter = binChunks.begin();
\r
1010 ChunkVector::const_iterator chunkEnd = binChunks.end();
\r
1011 for ( ; chunkIter != chunkEnd; ++chunkIter ) {
\r
1013 // get current chunk data
\r
1014 const Chunk& chunk = (*chunkIter);
\r
1015 const uint64_t& start = chunk.Start;
\r
1016 const uint64_t& stop = chunk.Stop;
\r
1018 // save chunk offsets
\r
1019 fwrite(&start, 8, 1, indexStream);
\r
1020 fwrite(&stop, 8, 1, indexStream);
\r
1024 // write linear offsets size
\r
1025 int32_t offsetSize = offsets.size();
\r
1026 fwrite(&offsetSize, 4, 1, indexStream);
\r
1028 // iterate over linear offsets
\r
1029 LinearOffsetVector::const_iterator offsetIter = offsets.begin();
\r
1030 LinearOffsetVector::const_iterator offsetEnd = offsets.end();
\r
1031 for ( ; offsetIter != offsetEnd; ++offsetIter ) {
\r
1033 // write linear offset value
\r
1034 const uint64_t& linearOffset = (*offsetIter);
\r
1035 fwrite(&linearOffset, 8, 1, indexStream);
\r
1039 // flush buffer, close file, and return success
\r
1040 fflush(indexStream);
\r
1041 fclose(indexStream);
\r