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: 29 March 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
44 // user-specified region values
\r
45 bool IsRegionSpecified;
\r
49 // BAM character constants
\r
50 const char* DNA_LOOKUP;
\r
51 const char* CIGAR_LOOKUP;
\r
53 // -------------------------------
\r
54 // constructor & destructor
\r
55 // -------------------------------
\r
56 BamReaderPrivate(void);
\r
57 ~BamReaderPrivate(void);
\r
59 // -------------------------------
\r
60 // "public" interface
\r
61 // -------------------------------
\r
65 bool Jump(int refID, int position = 0);
\r
66 void Open(const string& filename, const string& indexFilename = "");
\r
69 // access alignment data
\r
70 bool GetNextAlignment(BamAlignment& bAlignment);
\r
72 // access auxiliary data
\r
73 const string GetHeaderText(void) const;
\r
74 const int GetReferenceCount(void) const;
\r
75 const RefVector GetReferenceData(void) const;
\r
76 const int GetReferenceID(const string& refName) const;
\r
79 bool CreateIndex(void);
\r
81 // -------------------------------
\r
83 // -------------------------------
\r
85 // *** reading alignments and auxiliary data *** //
\r
87 // calculate bins that overlap region ( left to reference end for now )
\r
88 int BinsFromRegion(int refID, int left, uint16_t[MAX_BIN]);
\r
89 // calculates alignment end position based on starting position and provided CIGAR operations
\r
90 int CalculateAlignmentEnd(const int& position, const std::vector<CigarOp>& cigarData);
\r
91 // calculate file offset for first alignment chunk overlapping 'left'
\r
92 int64_t GetOffset(int refID, int left);
\r
93 // checks to see if alignment overlaps current region
\r
94 bool IsOverlap(BamAlignment& bAlignment);
\r
95 // retrieves header text from BAM file
\r
96 void LoadHeaderData(void);
\r
97 // retrieves BAM alignment under file pointer
\r
98 bool LoadNextAlignment(BamAlignment& bAlignment);
\r
99 // builds reference data structure from BAM file
\r
100 void LoadReferenceData(void);
\r
102 // *** index file handling *** //
\r
104 // calculates index for BAM file
\r
105 bool BuildIndex(void);
\r
106 // clear out inernal index data structure
\r
107 void ClearIndex(void);
\r
108 // saves BAM bin entry for index
\r
109 void InsertBinEntry(BamBinMap& binMap, const uint32_t& saveBin, const uint64_t& saveOffset, const uint64_t& lastOffset);
\r
110 // saves linear offset entry for index
\r
111 void InsertLinearOffset(LinearOffsetVector& offsets, const BamAlignment& bAlignment, const uint64_t& lastOffset);
\r
112 // loads index from BAM index file
\r
113 bool LoadIndex(void);
\r
114 // simplifies index by merging 'chunks'
\r
115 void MergeChunks(void);
\r
116 // round-up 32-bit integer to next power-of-2
\r
117 void Roundup32(int& value);
\r
118 // saves index to BAM index file
\r
119 bool WriteIndex(void);
\r
122 // -----------------------------------------------------
\r
123 // BamReader implementation (wrapper around BRPrivate)
\r
124 // -----------------------------------------------------
\r
127 BamReader::BamReader(void) {
\r
128 d = new BamReaderPrivate;
\r
132 BamReader::~BamReader(void) {
\r
138 void BamReader::Close(void) { d->Close(); }
\r
139 bool BamReader::Jump(int refID, int position) { return d->Jump(refID, position); }
\r
140 void BamReader::Open(const string& filename, const string& indexFilename) { d->Open(filename, indexFilename); }
\r
141 bool BamReader::Rewind(void) { return d->Rewind(); }
\r
143 // access alignment data
\r
144 bool BamReader::GetNextAlignment(BamAlignment& bAlignment) { return d->GetNextAlignment(bAlignment); }
\r
146 // access auxiliary data
\r
147 const string BamReader::GetHeaderText(void) const { return d->HeaderText; }
\r
148 const int BamReader::GetReferenceCount(void) const { return d->References.size(); }
\r
149 const RefVector BamReader::GetReferenceData(void) const { return d->References; }
\r
150 const int BamReader::GetReferenceID(const string& refName) const { return d->GetReferenceID(refName); }
\r
152 // index operations
\r
153 bool BamReader::CreateIndex(void) { return d->CreateIndex(); }
\r
155 // -----------------------------------------------------
\r
156 // BamReaderPrivate implementation
\r
157 // -----------------------------------------------------
\r
160 BamReader::BamReaderPrivate::BamReaderPrivate(void)
\r
161 : IsIndexLoaded(false)
\r
162 , AlignmentsBeginOffset(0)
\r
163 , IsRegionSpecified(false)
\r
166 , DNA_LOOKUP("=ACMGRSVTWYHKDBN")
\r
167 , CIGAR_LOOKUP("MIDNSHP")
\r
169 IsBigEndian = SystemIsBigEndian();
\r
173 BamReader::BamReaderPrivate::~BamReaderPrivate(void) {
\r
177 // calculate bins that overlap region ( left to reference end for now )
\r
178 int BamReader::BamReaderPrivate::BinsFromRegion(int refID, int left, uint16_t list[MAX_BIN]) {
\r
180 // get region boundaries
\r
181 uint32_t begin = (unsigned int)left;
\r
182 uint32_t end = (unsigned int)References.at(refID).RefLength - 1;
\r
184 // initialize list, bin '0' always a valid bin
\r
188 // get rest of bins that contain this region
\r
190 for (k = 1 + (begin>>26); k <= 1 + (end>>26); ++k) { list[i++] = k; }
\r
191 for (k = 9 + (begin>>23); k <= 9 + (end>>23); ++k) { list[i++] = k; }
\r
192 for (k = 73 + (begin>>20); k <= 73 + (end>>20); ++k) { list[i++] = k; }
\r
193 for (k = 585 + (begin>>17); k <= 585 + (end>>17); ++k) { list[i++] = k; }
\r
194 for (k = 4681 + (begin>>14); k <= 4681 + (end>>14); ++k) { list[i++] = k; }
\r
196 // return number of bins stored
\r
200 // populates BAM index data structure from BAM file data
\r
201 bool BamReader::BamReaderPrivate::BuildIndex(void) {
\r
203 // check to be sure file is open
\r
204 if (!mBGZF.IsOpen) { return false; }
\r
206 // move file pointer to beginning of alignments
\r
209 // get reference count, reserve index space
\r
210 int numReferences = References.size();
\r
211 for ( int i = 0; i < numReferences; ++i ) {
\r
212 Index.push_back(ReferenceIndex());
\r
215 // sets default constant for bin, ID, offset, coordinate variables
\r
216 const uint32_t defaultValue = 0xffffffffu;
\r
219 uint32_t saveBin(defaultValue);
\r
220 uint32_t lastBin(defaultValue);
\r
222 // reference ID data
\r
223 int32_t saveRefID(defaultValue);
\r
224 int32_t lastRefID(defaultValue);
\r
227 uint64_t saveOffset = mBGZF.Tell();
\r
228 uint64_t lastOffset = saveOffset;
\r
231 int32_t lastCoordinate = defaultValue;
\r
233 BamAlignment bAlignment;
\r
234 while( GetNextAlignment(bAlignment) ) {
\r
236 // change of chromosome, save ID, reset bin
\r
237 if ( lastRefID != bAlignment.RefID ) {
\r
238 lastRefID = bAlignment.RefID;
\r
239 lastBin = defaultValue;
\r
242 // if lastCoordinate greater than BAM position - file not sorted properly
\r
243 else if ( lastCoordinate > bAlignment.Position ) {
\r
244 printf("BAM file not properly sorted:\n");
\r
245 printf("Alignment %s : %d > %d on reference (id = %d)", bAlignment.Name.c_str(), lastCoordinate, bAlignment.Position, bAlignment.RefID);
\r
249 // if valid reference && BAM bin spans some minimum cutoff (smaller bin ids span larger regions)
\r
250 if ( (bAlignment.RefID >= 0) && (bAlignment.Bin < 4681) ) {
\r
252 // save linear offset entry (matched to BAM entry refID)
\r
253 ReferenceIndex& refIndex = Index.at(bAlignment.RefID);
\r
254 LinearOffsetVector& offsets = refIndex.Offsets;
\r
255 InsertLinearOffset(offsets, bAlignment, lastOffset);
\r
258 // if current BamAlignment bin != lastBin, "then possibly write the binning index"
\r
259 if ( bAlignment.Bin != lastBin ) {
\r
261 // if not first time through
\r
262 if ( saveBin != defaultValue ) {
\r
264 // save Bam bin entry
\r
265 ReferenceIndex& refIndex = Index.at(saveRefID);
\r
266 BamBinMap& binMap = refIndex.Bins;
\r
267 InsertBinEntry(binMap, saveBin, saveOffset, lastOffset);
\r
270 // update saveOffset
\r
271 saveOffset = lastOffset;
\r
273 // update bin values
\r
274 saveBin = bAlignment.Bin;
\r
275 lastBin = bAlignment.Bin;
\r
277 // update saveRefID
\r
278 saveRefID = bAlignment.RefID;
\r
280 // if invalid RefID, break out (why?)
\r
281 if ( saveRefID < 0 ) { break; }
\r
284 // make sure that current file pointer is beyond lastOffset
\r
285 if ( mBGZF.Tell() <= (int64_t)lastOffset ) {
\r
286 printf("Error in BGZF offsets.\n");
\r
290 // update lastOffset
\r
291 lastOffset = mBGZF.Tell();
\r
293 // update lastCoordinate
\r
294 lastCoordinate = bAlignment.Position;
\r
297 // save any leftover BAM data (as long as refID is valid)
\r
298 if ( saveRefID >= 0 ) {
\r
299 // save Bam bin entry
\r
300 ReferenceIndex& refIndex = Index.at(saveRefID);
\r
301 BamBinMap& binMap = refIndex.Bins;
\r
302 InsertBinEntry(binMap, saveBin, saveOffset, lastOffset);
\r
305 // simplify index by merging chunks
\r
308 // iterate over references
\r
309 BamIndex::iterator indexIter = Index.begin();
\r
310 BamIndex::iterator indexEnd = Index.end();
\r
311 for ( int i = 0; indexIter != indexEnd; ++indexIter, ++i ) {
\r
313 // get reference index data
\r
314 ReferenceIndex& refIndex = (*indexIter);
\r
315 BamBinMap& binMap = refIndex.Bins;
\r
316 LinearOffsetVector& offsets = refIndex.Offsets;
\r
318 // store whether reference has alignments or no
\r
319 References[i].RefHasAlignments = ( binMap.size() > 0 );
\r
321 // sort linear offsets
\r
322 sort(offsets.begin(), offsets.end());
\r
326 // rewind file pointer to beginning of alignments, return success/fail
\r
330 // calculates alignment end position based on starting position and provided CIGAR operations
\r
331 int BamReader::BamReaderPrivate::CalculateAlignmentEnd(const int& position, const vector<CigarOp>& cigarData) {
\r
333 // initialize alignment end to starting position
\r
334 int alignEnd = position;
\r
336 // iterate over cigar operations
\r
337 vector<CigarOp>::const_iterator cigarIter = cigarData.begin();
\r
338 vector<CigarOp>::const_iterator cigarEnd = cigarData.end();
\r
339 for ( ; cigarIter != cigarEnd; ++cigarIter) {
\r
340 char cigarType = (*cigarIter).Type;
\r
341 if ( cigarType == 'M' || cigarType == 'D' || cigarType == 'N' ) {
\r
342 alignEnd += (*cigarIter).Length;
\r
349 // clear index data structure
\r
350 void BamReader::BamReaderPrivate::ClearIndex(void) {
\r
351 Index.clear(); // sufficient ??
\r
354 // closes the BAM file
\r
355 void BamReader::BamReaderPrivate::Close(void) {
\r
358 HeaderText.clear();
\r
359 IsRegionSpecified = false;
\r
362 // create BAM index from BAM file (keep structure in memory) and write to default index output file
\r
363 bool BamReader::BamReaderPrivate::CreateIndex(void) {
\r
368 // build (& save) index from BAM file
\r
370 ok &= BuildIndex();
\r
371 ok &= WriteIndex();
\r
373 // return success/fail
\r
377 // returns RefID for given RefName (returns References.size() if not found)
\r
378 const int BamReader::BamReaderPrivate::GetReferenceID(const string& refName) const {
\r
380 // retrieve names from reference data
\r
381 vector<string> refNames;
\r
382 RefVector::const_iterator refIter = References.begin();
\r
383 RefVector::const_iterator refEnd = References.end();
\r
384 for ( ; refIter != refEnd; ++refIter) {
\r
385 refNames.push_back( (*refIter).RefName );
\r
388 // return 'index-of' refName ( if not found, returns refNames.size() )
\r
389 return distance(refNames.begin(), find(refNames.begin(), refNames.end(), refName));
\r
392 // get next alignment (from specified region, if given)
\r
393 bool BamReader::BamReaderPrivate::GetNextAlignment(BamAlignment& bAlignment) {
\r
395 // if valid alignment available
\r
396 if ( LoadNextAlignment(bAlignment) ) {
\r
398 // if region not specified, return success
\r
399 if ( !IsRegionSpecified ) { return true; }
\r
401 // load next alignment until region overlap is found
\r
402 while ( !IsOverlap(bAlignment) ) {
\r
403 // if no valid alignment available (likely EOF) return failure
\r
404 if ( !LoadNextAlignment(bAlignment) ) { return false; }
\r
407 // return success (alignment found that overlaps region)
\r
411 // no valid alignment
\r
412 else { return false; }
\r
415 // calculate closest indexed file offset for region specified
\r
416 int64_t BamReader::BamReaderPrivate::GetOffset(int refID, int left) {
\r
418 // calculate which bins overlap this region
\r
419 uint16_t* bins = (uint16_t*)calloc(MAX_BIN, 2);
\r
420 int numBins = BinsFromRegion(refID, left, bins);
\r
422 // get bins for this reference
\r
423 const ReferenceIndex& refIndex = Index.at(refID);
\r
424 const BamBinMap& binMap = refIndex.Bins;
\r
426 // get minimum offset to consider
\r
427 const LinearOffsetVector& offsets = refIndex.Offsets;
\r
428 uint64_t minOffset = ( (unsigned int)(left>>BAM_LIDX_SHIFT) >= offsets.size() ) ? 0 : offsets.at(left>>BAM_LIDX_SHIFT);
\r
430 // store offsets to beginning of alignment 'chunks'
\r
431 std::vector<int64_t> chunkStarts;
\r
433 // store all alignment 'chunk' starts for bins in this region
\r
434 for (int i = 0; i < numBins; ++i ) {
\r
435 uint16_t binKey = bins[i];
\r
437 map<uint32_t, ChunkVector>::const_iterator binIter = binMap.find(binKey);
\r
438 if ( (binIter != binMap.end()) && ((*binIter).first == binKey) ) {
\r
440 const ChunkVector& chunks = (*binIter).second;
\r
441 std::vector<Chunk>::const_iterator chunksIter = chunks.begin();
\r
442 std::vector<Chunk>::const_iterator chunksEnd = chunks.end();
\r
443 for ( ; chunksIter != chunksEnd; ++chunksIter) {
\r
444 const Chunk& chunk = (*chunksIter);
\r
445 if ( chunk.Stop > minOffset ) {
\r
446 chunkStarts.push_back( chunk.Start );
\r
455 // if no alignments found, else return smallest offset for alignment starts
\r
456 if ( chunkStarts.size() == 0 ) { return -1; }
\r
457 else { return *min_element(chunkStarts.begin(), chunkStarts.end()); }
\r
460 // saves BAM bin entry for index
\r
461 void BamReader::BamReaderPrivate::InsertBinEntry(BamBinMap& binMap,
\r
462 const uint32_t& saveBin,
\r
463 const uint64_t& saveOffset,
\r
464 const uint64_t& lastOffset)
\r
467 BamBinMap::iterator binIter = binMap.find(saveBin);
\r
469 // create new chunk
\r
470 Chunk newChunk(saveOffset, lastOffset);
\r
472 // if entry doesn't exist
\r
473 if ( binIter == binMap.end() ) {
\r
474 ChunkVector newChunks;
\r
475 newChunks.push_back(newChunk);
\r
476 binMap.insert( pair<uint32_t, ChunkVector>(saveBin, newChunks));
\r
481 ChunkVector& binChunks = (*binIter).second;
\r
482 binChunks.push_back( newChunk );
\r
486 // saves linear offset entry for index
\r
487 void BamReader::BamReaderPrivate::InsertLinearOffset(LinearOffsetVector& offsets,
\r
488 const BamAlignment& bAlignment,
\r
489 const uint64_t& lastOffset)
\r
491 // get converted offsets
\r
492 int beginOffset = bAlignment.Position >> BAM_LIDX_SHIFT;
\r
493 int endOffset = ( CalculateAlignmentEnd(bAlignment.Position, bAlignment.CigarData) - 1) >> BAM_LIDX_SHIFT;
\r
495 // resize vector if necessary
\r
496 int oldSize = offsets.size();
\r
497 int newSize = endOffset + 1;
\r
498 if ( oldSize < newSize ) {
\r
499 Roundup32(newSize);
\r
500 offsets.resize(newSize, 0);
\r
504 for(int i = beginOffset + 1; i <= endOffset ; ++i) {
\r
505 if ( offsets[i] == 0) {
\r
506 offsets[i] = lastOffset;
\r
511 // returns whether alignment overlaps currently specified region (refID, leftBound)
\r
512 bool BamReader::BamReaderPrivate::IsOverlap(BamAlignment& bAlignment) {
\r
514 // if on different reference sequence, quit
\r
515 if ( bAlignment.RefID != CurrentRefID ) { return false; }
\r
517 // read starts after left boundary
\r
518 if ( bAlignment.Position >= CurrentLeft) { return true; }
\r
520 // return whether alignment end overlaps left boundary
\r
521 return ( CalculateAlignmentEnd(bAlignment.Position, bAlignment.CigarData) >= CurrentLeft );
\r
524 // jumps to specified region(refID, leftBound) in BAM file, returns success/fail
\r
525 bool BamReader::BamReaderPrivate::Jump(int refID, int position) {
\r
527 // if data exists for this reference and position is valid
\r
528 if ( References.at(refID).RefHasAlignments && (position <= References.at(refID).RefLength) ) {
\r
530 // set current region
\r
531 CurrentRefID = refID;
\r
532 CurrentLeft = position;
\r
533 IsRegionSpecified = true;
\r
535 // calculate offset
\r
536 int64_t offset = GetOffset(CurrentRefID, CurrentLeft);
\r
538 // if in valid offset, return failure
\r
539 if ( offset == -1 ) { return false; }
\r
541 // otherwise return success of seek operation
\r
542 else { return mBGZF.Seek(offset); }
\r
545 // invalid jump request parameters, return failure
\r
549 // load BAM header data
\r
550 void BamReader::BamReaderPrivate::LoadHeaderData(void) {
\r
552 // check to see if proper BAM header
\r
554 if (mBGZF.Read(buffer, 4) != 4) {
\r
555 printf("Could not read header type\n");
\r
559 if (strncmp(buffer, "BAM\001", 4)) {
\r
560 printf("wrong header type!\n");
\r
564 // get BAM header text length
\r
565 mBGZF.Read(buffer, 4);
\r
566 unsigned int headerTextLength = BgzfData::UnpackUnsignedInt(buffer);
\r
567 if ( IsBigEndian ) { SwapEndian_32(headerTextLength); }
\r
569 // get BAM header text
\r
570 char* headerText = (char*)calloc(headerTextLength + 1, 1);
\r
571 mBGZF.Read(headerText, headerTextLength);
\r
572 HeaderText = (string)((const char*)headerText);
\r
574 // clean up calloc-ed temp variable
\r
578 // load existing index data from BAM index file (".bai"), return success/fail
\r
579 bool BamReader::BamReaderPrivate::LoadIndex(void) {
\r
581 // clear out index data
\r
584 // skip if index file empty
\r
585 if ( IndexFilename.empty() ) { return false; }
\r
587 // open index file, abort on error
\r
588 FILE* indexStream = fopen(IndexFilename.c_str(), "rb");
\r
590 printf("ERROR: Unable to open the BAM index file %s for reading.\n", IndexFilename.c_str() );
\r
594 size_t elementsRead = 0;
\r
596 // see if index is valid BAM index
\r
598 elementsRead = fread(magic, 1, 4, indexStream);
\r
599 if (strncmp(magic, "BAI\1", 4)) {
\r
600 printf("Problem with index file - invalid format.\n");
\r
601 fclose(indexStream);
\r
605 // get number of reference sequences
\r
606 uint32_t numRefSeqs;
\r
607 elementsRead = fread(&numRefSeqs, 4, 1, indexStream);
\r
608 if ( IsBigEndian ) { SwapEndian_32(numRefSeqs); }
\r
610 // intialize space for BamIndex data structure
\r
611 Index.reserve(numRefSeqs);
\r
613 // iterate over reference sequences
\r
614 for (unsigned int i = 0; i < numRefSeqs; ++i) {
\r
616 // get number of bins for this reference sequence
\r
618 elementsRead = fread(&numBins, 4, 1, indexStream);
\r
619 if ( IsBigEndian ) { SwapEndian_32(numBins); }
\r
622 RefData& refEntry = References[i];
\r
623 refEntry.RefHasAlignments = true;
\r
626 // intialize BinVector
\r
629 // iterate over bins for that reference sequence
\r
630 for (int j = 0; j < numBins; ++j) {
\r
634 elementsRead = fread(&binID, 4, 1, indexStream);
\r
636 // get number of regionChunks in this bin
\r
637 uint32_t numChunks;
\r
638 elementsRead = fread(&numChunks, 4, 1, indexStream);
\r
640 if ( IsBigEndian ) {
\r
641 SwapEndian_32(binID);
\r
642 SwapEndian_32(numChunks);
\r
645 // intialize ChunkVector
\r
646 ChunkVector regionChunks;
\r
647 regionChunks.reserve(numChunks);
\r
649 // iterate over regionChunks in this bin
\r
650 for (unsigned int k = 0; k < numChunks; ++k) {
\r
652 // get chunk boundaries (left, right)
\r
655 elementsRead = fread(&left, 8, 1, indexStream);
\r
656 elementsRead = fread(&right, 8, 1, indexStream);
\r
658 if ( IsBigEndian ) {
\r
659 SwapEndian_64(left);
\r
660 SwapEndian_64(right);
\r
664 regionChunks.push_back( Chunk(left, right) );
\r
667 // sort chunks for this bin
\r
668 sort( regionChunks.begin(), regionChunks.end(), ChunkLessThan );
\r
670 // save binID, chunkVector for this bin
\r
671 binMap.insert( pair<uint32_t, ChunkVector>(binID, regionChunks) );
\r
674 // load linear index for this reference sequence
\r
676 // get number of linear offsets
\r
677 int32_t numLinearOffsets;
\r
678 elementsRead = fread(&numLinearOffsets, 4, 1, indexStream);
\r
679 if ( IsBigEndian ) { SwapEndian_32(numLinearOffsets); }
\r
681 // intialize LinearOffsetVector
\r
682 LinearOffsetVector offsets;
\r
683 offsets.reserve(numLinearOffsets);
\r
685 // iterate over linear offsets for this reference sequeence
\r
686 uint64_t linearOffset;
\r
687 for (int j = 0; j < numLinearOffsets; ++j) {
\r
688 // read a linear offset & store
\r
689 elementsRead = fread(&linearOffset, 8, 1, indexStream);
\r
690 if ( IsBigEndian ) { SwapEndian_64(linearOffset); }
\r
691 offsets.push_back(linearOffset);
\r
694 // sort linear offsets
\r
695 sort( offsets.begin(), offsets.end() );
\r
697 // store index data for that reference sequence
\r
698 Index.push_back( ReferenceIndex(binMap, offsets) );
\r
701 // close index file (.bai) and return
\r
702 fclose(indexStream);
\r
706 // populates BamAlignment with alignment data under file pointer, returns success/fail
\r
707 bool BamReader::BamReaderPrivate::LoadNextAlignment(BamAlignment& bAlignment) {
\r
709 // read in the 'block length' value, make sure it's not zero
\r
711 mBGZF.Read(buffer, 4);
\r
712 unsigned int blockLength = BgzfData::UnpackUnsignedInt(buffer);
\r
713 if ( IsBigEndian ) { SwapEndian_32(blockLength); }
\r
714 if ( blockLength == 0 ) { return false; }
\r
716 // keep track of bytes read as method progresses
\r
719 // read in core alignment data, make sure the right size of data was read
\r
721 if ( mBGZF.Read(x, BAM_CORE_SIZE) != BAM_CORE_SIZE ) { return false; }
\r
722 bytesRead += BAM_CORE_SIZE;
\r
724 if ( IsBigEndian ) {
\r
725 for ( int i = 0; i < 8; ++i ) {
\r
726 SwapEndian_32(x[i]);
\r
730 // set BamAlignment 'core' data and character data lengths
\r
731 unsigned int tempValue;
\r
732 unsigned int queryNameLength;
\r
733 unsigned int numCigarOperations;
\r
734 unsigned int querySequenceLength;
\r
736 bAlignment.RefID = BgzfData::UnpackSignedInt(&x[0]);
\r
737 bAlignment.Position = BgzfData::UnpackSignedInt(&x[1]);
\r
739 tempValue = BgzfData::UnpackUnsignedInt(&x[2]);
\r
740 bAlignment.Bin = tempValue >> 16;
\r
741 bAlignment.MapQuality = tempValue >> 8 & 0xff;
\r
742 queryNameLength = tempValue & 0xff;
\r
744 tempValue = BgzfData::UnpackUnsignedInt(&x[3]);
\r
745 bAlignment.AlignmentFlag = tempValue >> 16;
\r
746 numCigarOperations = tempValue & 0xffff;
\r
748 querySequenceLength = BgzfData::UnpackUnsignedInt(&x[4]);
\r
749 bAlignment.MateRefID = BgzfData::UnpackSignedInt(&x[5]);
\r
750 bAlignment.MatePosition = BgzfData::UnpackSignedInt(&x[6]);
\r
751 bAlignment.InsertSize = BgzfData::UnpackSignedInt(&x[7]);
\r
753 // calculate lengths/offsets
\r
754 const unsigned int dataLength = blockLength - BAM_CORE_SIZE;
\r
755 const unsigned int cigarDataOffset = queryNameLength;
\r
756 const unsigned int seqDataOffset = cigarDataOffset + (numCigarOperations * 4);
\r
757 const unsigned int qualDataOffset = seqDataOffset + (querySequenceLength+1)/2;
\r
758 const unsigned int tagDataOffset = qualDataOffset + querySequenceLength;
\r
759 const unsigned int tagDataLen = dataLength - tagDataOffset;
\r
761 // set up destination buffers for character data
\r
762 char* allCharData = (char*)calloc(sizeof(char), dataLength);
\r
763 uint32_t* cigarData = (uint32_t*)(allCharData + cigarDataOffset);
\r
764 char* seqData = ((char*)allCharData) + seqDataOffset;
\r
765 char* qualData = ((char*)allCharData) + qualDataOffset;
\r
766 char* tagData = ((char*)allCharData) + tagDataOffset;
\r
768 // get character data - make sure proper data size was read
\r
769 if ( mBGZF.Read(allCharData, dataLength) != (signed int)dataLength) { return false; }
\r
772 bytesRead += dataLength;
\r
774 // clear out any previous string data
\r
775 bAlignment.Name.clear(;)
\r
776 bAlignment.QueryBases.clear();
\r
777 bAlignment.Qualities.clear();
\r
778 bAlignment.AlignedBases.clear();
\r
779 bAlignment.CigarData.clear();
\r
780 bAlignment.TagData.clear();
\r
783 bAlignment.Name = (string)((const char*)(allCharData));
\r
785 // save query sequence
\r
786 // -----------------------
\r
787 // Added: 3-25-2010 DWB
\r
788 // Improved: reduced repeated memory allocations as string grows
\r
789 bAlignment.QueryBases.reserve(querySequenceLength);
\r
790 // -----------------------
\r
792 for (unsigned int i = 0; i < querySequenceLength; ++i) {
\r
793 char singleBase = DNA_LOOKUP[ ( ( seqData[(i/2)] >> (4*(1-(i%2)))) & 0xf ) ];
\r
794 bAlignment.QueryBases.append( 1, singleBase );
\r
797 // save sequence length
\r
798 bAlignment.Length = bAlignment.QueryBases.length();
\r
800 // save qualities, convert from numeric QV to FASTQ character
\r
801 // -----------------------
\r
802 // Added: 3-25-2010 DWB
\r
803 // Improved: reduced repeated memory allocations as string grows
\r
804 bAlignment.Qualities.reserve(querySequenceLength);
\r
805 // -----------------------
\r
807 for (unsigned int i = 0; i < querySequenceLength; ++i) {
\r
808 char singleQuality = (char)(qualData[i]+33);
\r
809 bAlignment.Qualities.append( 1, singleQuality );
\r
812 // save CIGAR-related data;
\r
813 // -----------------------
\r
814 // Added: 3-25-2010 DWB
\r
815 // Improved: reduced repeated memory allocations as string grows
\r
816 bAlignment.AlignedBases.reserve(querySequenceLength);
\r
817 // -----------------------
\r
820 for (unsigned int i = 0; i < numCigarOperations; ++i) {
\r
822 if ( IsBigEndian ) { SwapEndian_32(cigarData[i]); }
\r
824 // build CigarOp struct
\r
826 op.Length = (cigarData[i] >> BAM_CIGAR_SHIFT);
\r
827 op.Type = CIGAR_LOOKUP[ (cigarData[i] & BAM_CIGAR_MASK) ];
\r
830 bAlignment.CigarData.push_back(op);
\r
832 // build AlignedBases string
\r
837 bAlignment.AlignedBases.append( bAlignment.QueryBases.substr(k, op.Length) ); // for 'M', 'I' - write bases
\r
841 k += op.Length; // for 'S' - skip over query bases
\r
845 bAlignment.AlignedBases.append( op.Length, '-' ); // for 'D' - write gap character
\r
849 bAlignment.AlignedBases.append( op.Length, '*' ); // for 'P' - write padding character;
\r
853 bAlignment.AlignedBases.append( op.Length, 'N' ); // for 'N' - write N's, skip bases in query sequence
\r
854 // -----------------------
\r
855 // Removed: 3-25-2010 DWB
\r
856 // Contributed: ARQ
\r
857 // Fixed: compliance with actual 'N' definition in BAM spec
\r
859 // -----------------------
\r
863 break; // for 'H' - do nothing, move to next op
\r
866 printf("ERROR: Invalid Cigar op type\n"); // shouldn't get here
\r
872 // -----------------------
\r
873 // Added: 3-25-2010 DWB
\r
874 // Fixed: endian-correctness for tag data
\r
875 // -----------------------
\r
876 if ( IsBigEndian ) {
\r
878 while ( i < tagDataLen ) {
\r
880 i += 2; // skip tag type (e.g. "RG", "NM", etc)
\r
881 uint8_t type = toupper(tagData[i]); // lower & upper case letters have same meaning
\r
882 ++i; // skip value type
\r
892 SwapEndian_16p(&tagData[i]);
\r
893 i+=2; // sizeof(uint16_t)
\r
898 SwapEndian_32p(&tagData[i]);
\r
899 i+=4; // sizeof(uint32_t)
\r
903 SwapEndian_64p(&tagData[i]);
\r
904 i+=8; // sizeof(uint64_t)
\r
909 while (tagData[i]) { ++i; }
\r
910 ++i; // increment one more for null terminator
\r
914 printf("ERROR: Invalid tag value type\n"); // shouldn't get here
\r
922 bAlignment.TagData.resize(tagDataLen);
\r
923 memcpy((char*)bAlignment.TagData.data(), tagData, tagDataLen);
\r
930 // loads reference data from BAM file
\r
931 void BamReader::BamReaderPrivate::LoadReferenceData(void) {
\r
933 // get number of reference sequences
\r
935 mBGZF.Read(buffer, 4);
\r
936 unsigned int numberRefSeqs = BgzfData::UnpackUnsignedInt(buffer);
\r
937 if ( IsBigEndian ) { SwapEndian_32(numberRefSeqs); }
\r
938 if (numberRefSeqs == 0) { return; }
\r
939 References.reserve((int)numberRefSeqs);
\r
941 // iterate over all references in header
\r
942 for (unsigned int i = 0; i != numberRefSeqs; ++i) {
\r
944 // get length of reference name
\r
945 mBGZF.Read(buffer, 4);
\r
946 unsigned int refNameLength = BgzfData::UnpackUnsignedInt(buffer);
\r
947 if ( IsBigEndian ) { SwapEndian_32(refNameLength); }
\r
948 char* refName = (char*)calloc(refNameLength, 1);
\r
950 // get reference name and reference sequence length
\r
951 mBGZF.Read(refName, refNameLength);
\r
952 mBGZF.Read(buffer, 4);
\r
953 int refLength = BgzfData::UnpackSignedInt(buffer);
\r
954 if ( IsBigEndian ) { SwapEndian_32(refLength); }
\r
956 // store data for reference
\r
957 RefData aReference;
\r
958 aReference.RefName = (string)((const char*)refName);
\r
959 aReference.RefLength = refLength;
\r
960 References.push_back(aReference);
\r
962 // clean up calloc-ed temp variable
\r
967 // merges 'alignment chunks' in BAM bin (used for index building)
\r
968 void BamReader::BamReaderPrivate::MergeChunks(void) {
\r
970 // iterate over reference enties
\r
971 BamIndex::iterator indexIter = Index.begin();
\r
972 BamIndex::iterator indexEnd = Index.end();
\r
973 for ( ; indexIter != indexEnd; ++indexIter ) {
\r
975 // get BAM bin map for this reference
\r
976 ReferenceIndex& refIndex = (*indexIter);
\r
977 BamBinMap& bamBinMap = refIndex.Bins;
\r
979 // iterate over BAM bins
\r
980 BamBinMap::iterator binIter = bamBinMap.begin();
\r
981 BamBinMap::iterator binEnd = bamBinMap.end();
\r
982 for ( ; binIter != binEnd; ++binIter ) {
\r
984 // get chunk vector for this bin
\r
985 ChunkVector& binChunks = (*binIter).second;
\r
986 if ( binChunks.size() == 0 ) { continue; }
\r
988 ChunkVector mergedChunks;
\r
989 mergedChunks.push_back( binChunks[0] );
\r
991 // iterate over chunks
\r
993 ChunkVector::iterator chunkIter = binChunks.begin();
\r
994 ChunkVector::iterator chunkEnd = binChunks.end();
\r
995 for ( ++chunkIter; chunkIter != chunkEnd; ++chunkIter) {
\r
997 // get 'currentChunk' based on numeric index
\r
998 Chunk& currentChunk = mergedChunks[i];
\r
1000 // get iteratorChunk based on vector iterator
\r
1001 Chunk& iteratorChunk = (*chunkIter);
\r
1003 // if currentChunk.Stop(shifted) == iterator Chunk.Start(shifted)
\r
1004 if ( currentChunk.Stop>>16 == iteratorChunk.Start>>16 ) {
\r
1006 // set currentChunk.Stop to iteratorChunk.Stop
\r
1007 currentChunk.Stop = iteratorChunk.Stop;
\r
1012 // set currentChunk + 1 to iteratorChunk
\r
1013 mergedChunks.push_back(iteratorChunk);
\r
1018 // saved merged chunk vector
\r
1019 (*binIter).second = mergedChunks;
\r
1024 // opens BAM file (and index)
\r
1025 void BamReader::BamReaderPrivate::Open(const string& filename, const string& indexFilename) {
\r
1027 Filename = filename;
\r
1028 IndexFilename = indexFilename;
\r
1030 // open the BGZF file for reading, retrieve header text & reference data
\r
1031 mBGZF.Open(filename, "rb");
\r
1033 LoadReferenceData();
\r
1035 // store file offset of first alignment
\r
1036 AlignmentsBeginOffset = mBGZF.Tell();
\r
1038 // open index file & load index data (if exists)
\r
1039 if ( !IndexFilename.empty() ) {
\r
1044 // returns BAM file pointer to beginning of alignment data
\r
1045 bool BamReader::BamReaderPrivate::Rewind(void) {
\r
1047 // find first reference that has alignments in the BAM file
\r
1049 int refCount = References.size();
\r
1050 for ( ; refID < refCount; ++refID ) {
\r
1051 if ( References.at(refID).RefHasAlignments ) { break; }
\r
1054 // store default bounds for first alignment
\r
1055 CurrentRefID = refID;
\r
1057 IsRegionSpecified = false;
\r
1059 // return success/failure of seek
\r
1060 return mBGZF.Seek(AlignmentsBeginOffset);
\r
1063 // rounds value up to next power-of-2 (used in index building)
\r
1064 void BamReader::BamReaderPrivate::Roundup32(int& value) {
\r
1066 value |= value >> 1;
\r
1067 value |= value >> 2;
\r
1068 value |= value >> 4;
\r
1069 value |= value >> 8;
\r
1070 value |= value >> 16;
\r
1074 // saves index data to BAM index file (".bai"), returns success/fail
\r
1075 bool BamReader::BamReaderPrivate::WriteIndex(void) {
\r
1077 IndexFilename = Filename + ".bai";
\r
1078 FILE* indexStream = fopen(IndexFilename.c_str(), "wb");
\r
1079 if ( indexStream == 0 ) {
\r
1080 printf("ERROR: Could not open file to save index\n");
\r
1084 // write BAM index header
\r
1085 fwrite("BAI\1", 1, 4, indexStream);
\r
1087 // write number of reference sequences
\r
1088 int32_t numReferenceSeqs = Index.size();
\r
1089 if ( IsBigEndian ) { SwapEndian_32(numReferenceSeqs); }
\r
1090 fwrite(&numReferenceSeqs, 4, 1, indexStream);
\r
1092 // iterate over reference sequences
\r
1093 BamIndex::const_iterator indexIter = Index.begin();
\r
1094 BamIndex::const_iterator indexEnd = Index.end();
\r
1095 for ( ; indexIter != indexEnd; ++ indexIter ) {
\r
1097 // get reference index data
\r
1098 const ReferenceIndex& refIndex = (*indexIter);
\r
1099 const BamBinMap& binMap = refIndex.Bins;
\r
1100 const LinearOffsetVector& offsets = refIndex.Offsets;
\r
1102 // write number of bins
\r
1103 int32_t binCount = binMap.size();
\r
1104 if ( IsBigEndian ) { SwapEndian_32(binCount); }
\r
1105 fwrite(&binCount, 4, 1, indexStream);
\r
1107 // iterate over bins
\r
1108 BamBinMap::const_iterator binIter = binMap.begin();
\r
1109 BamBinMap::const_iterator binEnd = binMap.end();
\r
1110 for ( ; binIter != binEnd; ++binIter ) {
\r
1112 // get bin data (key and chunk vector)
\r
1113 uint32_t binKey = (*binIter).first;
\r
1114 const ChunkVector& binChunks = (*binIter).second;
\r
1116 // save BAM bin key
\r
1117 if ( IsBigEndian ) { SwapEndian_32(binKey); }
\r
1118 fwrite(&binKey, 4, 1, indexStream);
\r
1120 // save chunk count
\r
1121 int32_t chunkCount = binChunks.size();
\r
1122 if ( IsBigEndian ) { SwapEndian_32(chunkCount); }
\r
1123 fwrite(&chunkCount, 4, 1, indexStream);
\r
1125 // iterate over chunks
\r
1126 ChunkVector::const_iterator chunkIter = binChunks.begin();
\r
1127 ChunkVector::const_iterator chunkEnd = binChunks.end();
\r
1128 for ( ; chunkIter != chunkEnd; ++chunkIter ) {
\r
1130 // get current chunk data
\r
1131 const Chunk& chunk = (*chunkIter);
\r
1132 uint64_t start = chunk.Start;
\r
1133 uint64_t stop = chunk.Stop;
\r
1135 if ( IsBigEndian ) {
\r
1136 SwapEndian_64(start);
\r
1137 SwapEndian_64(stop);
\r
1140 // save chunk offsets
\r
1141 fwrite(&start, 8, 1, indexStream);
\r
1142 fwrite(&stop, 8, 1, indexStream);
\r
1146 // write linear offsets size
\r
1147 int32_t offsetSize = offsets.size();
\r
1148 if ( IsBigEndian ) { SwapEndian_32(offsetSize); }
\r
1149 fwrite(&offsetSize, 4, 1, indexStream);
\r
1151 // iterate over linear offsets
\r
1152 LinearOffsetVector::const_iterator offsetIter = offsets.begin();
\r
1153 LinearOffsetVector::const_iterator offsetEnd = offsets.end();
\r
1154 for ( ; offsetIter != offsetEnd; ++offsetIter ) {
\r
1156 // write linear offset value
\r
1157 uint64_t linearOffset = (*offsetIter);
\r
1158 if ( IsBigEndian ) { SwapEndian_64(linearOffset); }
\r
1159 fwrite(&linearOffset, 8, 1, indexStream);
\r
1163 // flush buffer, close file, and return success
\r
1164 fflush(indexStream);
\r
1165 fclose(indexStream);
\r