C++ SamRecord::getMapQuality方法代码示例

void GenomeRegionSeqStats::CalcClusters(String &bamFile, int minMapQuality)
{
  SamFile sam;
  SamRecord samRecord;
  SamFileHeader samHeader;

  if(!sam.OpenForRead(bamFile.c_str()))
    error("Open BAM file %s failed!\n", bamFile.c_str());

  if(!sam.ReadHeader(samHeader)) {
      error("Read BAM file header %s failed!\n", bamFile.c_str());
  }
  
  if(depth.size()==0) depth.resize(referencegenome.sequenceLength());
  
  String contigLabel;
  uint32_t start;
  uint32_t gstart;
  Reset();
  while(sam.ReadRecord(samHeader, samRecord))
    {
      nReads++;
      if(samRecord.getFlag() & SamFlag::UNMAPPED) nUnMapped++;

      if(samRecord.getMapQuality() < minMapQuality) continue;

      CigarRoller cigar(samRecord.getCigar());

      int nonClipSequence = 0;

      if(cigar.size()!=0 && cigar[0].operation==Cigar::softClip)
          nonClipSequence = cigar[0].count;

      contigLabel = samRecord.getReferenceName();
      start = nonClipSequence + samRecord.get0BasedPosition();  // start is 0-based

      gstart = referencegenome.getGenomePosition(contigLabel.c_str(), start);

      if(IsInRegions(contigLabel, start, start+samRecord.getReadLength())) continue;

      for(uint32_t i=gstart; i<gstart+samRecord.getReadLength(); i++)
       if(depth[i]<MAXDP)
        depth[i]++;
      nMappedOutTargets++;
    }
}

int Likelihood::getFlankingCount( string & chr, int pos )
{
	bool status = bam.SetReadSection( chr.c_str(), pos - avr_read_length * 2, pos + avr_read_length/2 );
	if (!status)
		return 0;

	int n = 0;
	SamRecord rec;
	while(bam.ReadRecord(bam_header, rec)) {
		if ( !(rec.getFlag() & 0x2) )
			continue;
		if (IsSupplementary(rec.getFlag()))
			continue;
		if (rec.getReadLength() < avr_read_length / 3)
			continue;
		if (rec.getMapQuality() < MIN_QUALITY)
			continue;
		if (rec.get1BasedPosition() + MIN_CLIP/2 <= pos && rec.get1BasedAlignmentEnd() - MIN_CLIP/2 >= pos )
			n++;
	}

	return n;
}

bool Recab::processReadBuildTable(SamRecord& samRecord)
{
    static BaseData data;
    static std::string chromosomeName;
    static std::string readGroup;
    static std::string aligTypes;

    int seqLen = samRecord.getReadLength();
    
    // Check if the parameters have been processed.
    if(!myParamsSetup)
    {
        // This throws an exception if the reference cannot be setup.
        processParams();
    }

    uint16_t  flag = samRecord.getFlag();

    if(!SamFlag::isMapped(flag))
    {
        // Unmapped, skip processing
        ++myUnMappedCount;
    }
    else
    {
        // This read is mapped.
        ++myMappedCount;
    }

    if(SamFlag::isSecondary(flag))
    {
        // Secondary read
        ++mySecondaryCount;
    }
    if(SamFlag::isDuplicate(flag))
    {
        ++myDupCount;
    }
    if(SamFlag::isQCFailure(flag))
    {
        ++myQCFailCount;
    }

    // Check if the flag contains an exclude.
    if((flag & myIntBuildExcludeFlags) != 0)
    {
        // Do not use this read for building the recalibration table.
        ++myNumBuildSkipped;
        return(false);
    }

    if(samRecord.getMapQuality() == 0)
    {
        // 0 mapping quality, so skip processing.
        ++myMapQual0Count;
        ++myNumBuildSkipped;
        return(false);
    }
    if(samRecord.getMapQuality() == 255)
    {
        // 255 mapping quality, so skip processing.
        ++myMapQual255Count;
        ++myNumBuildSkipped;
        return(false);
    }
    
    chromosomeName = samRecord.getReferenceName();
    readGroup = samRecord.getString("RG").c_str();

    // Look for the read group in the map.
    // TODO - extra string constructor??
    RgInsertReturn insertRet = 
        myRg2Id.insert(std::pair<std::string, uint16_t>(readGroup, 0));
    if(insertRet.second == true)
    {
        // New element inserted.
        insertRet.first->second = myId2Rg.size();
        myId2Rg.push_back(readGroup);
    }

    data.rgid = insertRet.first->second;


    //reverse
    bool reverse;
    if(SamFlag::isReverse(flag))
        reverse = true;
    else
        reverse = false;

    if(myReferenceGenome == NULL)
    {
        throw std::runtime_error("Failed to setup Reference File.\n");
    }

    genomeIndex_t mapPos = 
        myReferenceGenome->getGenomePosition(chromosomeName.c_str(), 
                                             samRecord.get1BasedPosition());

    if(mapPos==INVALID_GENOME_INDEX)
//.........这里部分代码省略.........

void validateRead1ModQuality(SamRecord& samRecord)
{
    //////////////////////////////////////////
    // Validate Record 1
    // Create record structure for validating.
    int expectedBlockSize = 89;
    const char* expectedReferenceName = "1";
    const char* expectedMateReferenceName = "1";
    const char* expectedMateReferenceNameOrEqual = "=";

    bamRecordStruct* expectedRecordPtr =
        (bamRecordStruct *) malloc(expectedBlockSize + sizeof(int));

    char tag[3];
    char type;
    void* value;
    bamRecordStruct* bufferPtr;
    unsigned char* varPtr;

    expectedRecordPtr->myBlockSize = expectedBlockSize;
    expectedRecordPtr->myReferenceID = 0;
    expectedRecordPtr->myPosition = 1010;
    expectedRecordPtr->myReadNameLength = 23;
    expectedRecordPtr->myMapQuality = 0;
    expectedRecordPtr->myBin = 4681;
    expectedRecordPtr->myCigarLength = 2;
    expectedRecordPtr->myFlag = 73;
    expectedRecordPtr->myReadLength = 5;
    expectedRecordPtr->myMateReferenceID = 0;
    expectedRecordPtr->myMatePosition = 1010;
    expectedRecordPtr->myInsertSize = 0;
   
    // Check the alignment end
    assert(samRecord.get0BasedAlignmentEnd() == 1016);
    assert(samRecord.get1BasedAlignmentEnd() == 1017);
    assert(samRecord.getAlignmentLength() == 7);
    assert(samRecord.get0BasedUnclippedStart() == 1010);
    assert(samRecord.get1BasedUnclippedStart() == 1011);
    assert(samRecord.get0BasedUnclippedEnd() == 1016);
    assert(samRecord.get1BasedUnclippedEnd() == 1017);

    // Check the accessors.
    assert(samRecord.getBlockSize() == expectedRecordPtr->myBlockSize);
    assert(samRecord.getReferenceID() == expectedRecordPtr->myReferenceID);
    assert(strcmp(samRecord.getReferenceName(), expectedReferenceName) == 0);
    assert(samRecord.get1BasedPosition() == expectedRecordPtr->myPosition + 1);
    assert(samRecord.get0BasedPosition() == expectedRecordPtr->myPosition);
    assert(samRecord.getReadNameLength() == 
           expectedRecordPtr->myReadNameLength);
    assert(samRecord.getMapQuality() == expectedRecordPtr->myMapQuality);
    assert(samRecord.getBin() == expectedRecordPtr->myBin);
    assert(samRecord.getCigarLength() == expectedRecordPtr->myCigarLength);
    assert(samRecord.getFlag() == expectedRecordPtr->myFlag);
    assert(samRecord.getReadLength() == expectedRecordPtr->myReadLength);
    assert(samRecord.getMateReferenceID() ==
           expectedRecordPtr->myMateReferenceID);
    assert(strcmp(samRecord.getMateReferenceName(), 
                  expectedMateReferenceName) == 0);
    assert(strcmp(samRecord.getMateReferenceNameOrEqual(), 
                  expectedMateReferenceNameOrEqual) == 0);
    assert(samRecord.get1BasedMatePosition() == 
           expectedRecordPtr->myMatePosition + 1);
    assert(samRecord.get0BasedMatePosition() ==
           expectedRecordPtr->myMatePosition);
    assert(samRecord.getInsertSize() == expectedRecordPtr->myInsertSize);
    assert(strcmp(samRecord.getReadName(), "1:1011:F:255+17M15D20M") == 0);
    assert(strcmp(samRecord.getCigar(), "5M2D") == 0);
    assert(strcmp(samRecord.getSequence(), "CCGAA") == 0);
    assert(strcmp(samRecord.getQuality(), "ABCDE") == 0);
    assert(samRecord.getNumOverlaps(1010, 1017) == 5);
    assert(samRecord.getNumOverlaps(1010, 1016) == 5);
    assert(samRecord.getNumOverlaps(1012, 1017) == 3);
    assert(samRecord.getNumOverlaps(1015, 1017) == 0);
    assert(samRecord.getNumOverlaps(1017, 1010) == 0);
    assert(samRecord.getNumOverlaps(1013, 1011) == 0);
    assert(samRecord.getNumOverlaps(-1, 1017) == 5);

    // Reset the tag iter, since the tags have already been read.
    samRecord.resetTagIter();

    // Check the tags.
    assert(samRecord.getNextSamTag(tag, type, &value) == true);
    assert(tag[0] == 'A');
    assert(tag[1] == 'M');
    assert(type == 'i');
    assert(*(char*)value == 0);
    assert(samRecord.getNextSamTag(tag, type, &value) == true);
    assert(tag[0] == 'M');
    assert(tag[1] == 'D');
    assert(type == 'Z');
    assert(*(String*)value == "37");
    assert(samRecord.getNextSamTag(tag, type, &value) == true);
    assert(tag[0] == 'N');
    assert(tag[1] == 'M');
    assert(type == 'i');
    assert(*(char*)value == 0);
    assert(samRecord.getNextSamTag(tag, type, &value) == true);
    assert(tag[0] == 'X');
    assert(tag[1] == 'T');
    assert(type == 'A');
//.........这里部分代码省略.........

// Add an entry to this pileup element.  
void PileupElementBaseQual::addEntry(SamRecord& record)
{
    // Call the base class:
    PileupElement::addEntry(record);

    if(myRefAllele.empty())
    {
    	genomeIndex_t markerIndex = (*myRefSeq).getGenomePosition(getChromosome(), static_cast<uint32_t>(getRefPosition()+1));
        myRefAllele = (*myRefSeq)[markerIndex];
    }

    // Increment the index
    ++myIndex;
    
    // if the index has gone beyond the allocated space, double the size.
    if(myIndex >= myAllocatedSize)
    {
        char* tempBuffer = (char*)realloc(myBases, myAllocatedSize * 2);
        if(tempBuffer == NULL)
        {
            std::cerr << "Memory Allocation Failure\n";
            // TODO
            return;
        }
        myBases = tempBuffer;
        int8_t* tempInt8Buffer = (int8_t*)realloc(myMapQualities, myAllocatedSize * 2 * sizeof(int8_t));
        if(tempInt8Buffer == NULL)
        {
            std::cerr << "Memory Allocation Failure\n";
            // TODO
            return;
        }
        myMapQualities = tempInt8Buffer; 
        tempInt8Buffer = (int8_t*)realloc(myQualities, myAllocatedSize * 2 * sizeof(int8_t));
        if(tempInt8Buffer == NULL)
        {
            std::cerr << "Memory Allocation Failure\n";
            // TODO
            return;
        }
        myQualities = tempInt8Buffer;
        tempBuffer = (char*)realloc(myStrands, myAllocatedSize * 2);
        if(tempBuffer == NULL)
        {
            std::cerr << "Memory Allocation Failure\n";
            // TODO
            return;
        }
        myStrands = tempBuffer;
        tempInt8Buffer = (int8_t*)realloc(myCycles, myAllocatedSize * 2 * sizeof(int8_t));
        if(tempInt8Buffer == NULL)
        {
            std::cerr << "Memory Allocation Failure\n";
            // TODO
            return;
        }
        myCycles = tempInt8Buffer; 
        int16_t* tempInt16Buffer = (int16_t*)realloc(myGLScores, myAllocatedSize * 2 * sizeof(int16_t));
        if(tempInt8Buffer == NULL)
        {
            std::cerr << "Memory Allocation Failure\n";
            // TODO
            return;
        }
        myGLScores = tempInt16Buffer;
        myAllocatedSize = myAllocatedSize * 2;
    }

    Cigar* cigar = record.getCigarInfo();
    
    if(cigar == NULL)
    {
        throw std::runtime_error("Failed to retrieve cigar info from the record.");
    }

    int32_t readIndex = 
        cigar->getQueryIndex(getRefPosition(), record.get0BasedPosition());

    // If the readPosition is N/A, this is a deletion.
    if(readIndex != CigarRoller::INDEX_NA)
    {
        char base = record.getSequence(readIndex);
        int8_t mapQual = record.getMapQuality();
        //-33 to obtain the PHRED base quality
        char qual = record.getQuality(readIndex) - 33;
        if(qual == UNSET_QUAL)
        {
            qual = ' ';
        }
        char strand = (record.getFlag() & 0x0010) ? 'R' : 'F';
        int cycle = strand == 'F' ? readIndex + 1 : record.getReadLength() -  readIndex;
        myBases[myIndex] = base;
        myMapQualities[myIndex] = mapQual;
        myQualities[myIndex] = qual;
        myStrands[myIndex] = strand;
        myCycles[myIndex] = cycle;
    }
    else if(myAddDelAsBase)
    {
//.........这里部分代码省略.........

/*
	if a discordant read is mapped to MEI (overlap with MEI coord)
		add centor of ( anchor_end + 3*avr_ins_var )
	skip unmap & clip
	check 3 types at the same time
*/
void Sites::AddDiscoverSiteFromSingleBam( SingleSampleInfo & si )
{
/*for(int i=0;i<NMEI; i++) {
std::cout << "m=" << i << ": ";
for(map<string, map<int, bool> >::iterator t=meiCoord[m].begin(); t!=meiCoord[m].end(); t++)
std::cout << t->first << "->" << t->second.size() << " ";
std::cout << std::endl;
}*/

	avr_read_length = si.avr_read_length;
	avr_ins_size = si.avr_ins_size;
	min_read_length = avr_read_length / 3;
	current_depth = si.depth;
//	total_depth += current_depth;

	resetDepthAddFlag();
	
	SamFile bam;
	SamFileHeader bam_header;
	OpenBamAndBai( bam,bam_header, si.bam_name );
	for( int i=0; i<pchr_list->size(); i++ ) {
		string chr = (*pchr_list)[i];
//		if ( !single_chr.empty() && chr.compare(single_chr)!=0 )
//			continue;
		if ( siteList.find(chr) == siteList.end() )
			siteList[chr].clear();
//		map<string, map<int, SingleSite> >::iterator pchr = siteList[m].find(chr);
//		map<string, map<int, bool> >::iterator coord_chr_ptr = meiCoord[m].find(chr);
//		if (coord_chr_ptr == meiCoord[m].end())
//			continue;
		bool section_status;
		if (range_start<0) { // no range
			section_status = bam.SetReadSection( chr.c_str() );
			if (!section_status) {
				string str = "Cannot set read section at chr " + chr;
				morphWarning( str );
			}
		}
		else { // set range
			section_status = bam.SetReadSection( chr.c_str(), range_start, range_end );
			if (!section_status) {
				string str = "Cannot set read section at chr " + chr + " " + std::to_string(range_start) + "-" + std::to_string(range_end); 
				morphWarning( str );
			}			
		}
		
		// DO ADDING
//		if (siteList[chr].empty())
//			p_reach_last = 1;
//		else {
//			p_reach_last = 0;
		pnearest = siteList[chr].begin();
//		}
		SingleSite new_site; // temporary cluster. will be added to map later.
		new_site.depth = current_depth;
		bool start_new = 1; // check if need to add new_site to map and start new new_site
		SamRecord rec;
		int between = 0; // count #reads after new_site.end. If end changed, add it to rcount and reset to zero
		while( bam.ReadRecord( bam_header, rec ) ) {
			if (!start_new) {
				if (rec.get1BasedPosition() >= new_site.end)
					between++;
				else
					new_site.rcount++;
			}
			if (rec.getFlag() & 0x2)
				continue;
			if ( OneEndUnmap( rec.getFlag() ) )
				continue;
			if ( IsSupplementary(rec.getFlag()) )
				continue;
			if ( rec.getReadLength() < min_read_length )
				continue;
			if ( rec.getMapQuality() < MIN_QUALITY )
				continue;
			if (chr.compare(rec.getMateReferenceName())==0 && rec.getInsertSize() < abs(avr_ins_size*2))
				continue;
			bool is_mei = 0;
			vector<bool> is_in_coord;
			is_in_coord.resize(3, 0);
			for(int m=0; m<NMEI; m++) {
				map<string, map<int, bool> >::iterator coord_chr_ptr = meiCoord[m].find(rec.getMateReferenceName());
				if (coord_chr_ptr == meiCoord[m].end())
					is_in_coord[m] = 0;
				else
					is_in_coord[m] = isWithinCoord( rec.get1BasedMatePosition(), coord_chr_ptr->second ); // within MEI coord
				if (is_in_coord[m])
					is_mei = 1;
			}
			if (!is_mei)
				continue;
			if (start_new) {
				setNewCluster( is_in_coord, new_site,rec);
				start_new = 0;
//.........这里部分代码省略.........