Java SAMRecord.getReadLength方法代码示例

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
/**
 * Converts a SAM record into a bed entry and outputs it straight away
 *
 * @param record
 */
private void processOneUnalignedRecord(SAMRecord record) {

    // output the alignment positions into a bed file
    int start = record.getAlignmentStart() - 1;
    int end = start + record.getReadLength();
    String out = record.getReferenceName() + "\t" + start + "\t" + end + "\n";

    synchronized (outunaligned) {
        try {
            outunaligned.write(out.getBytes());
        } catch (Exception ex) {
            System.out.println("Exception writing unaligned bed: " + ex.getMessage());
        }
    }

    // create a dummy thesaurus entry
    ThesaurusEntry entry = new ThesaurusEntry(record, ginfo);
    // complete it (need to fix alignment end manually, then use function to get anchors set right)
    entry.alignEnd = entry.originEnd;
    completeThesaurusEntry(entry, record);
    // set the origin chromosome to -1 to signal unaligned setup
    entry.originChrIndex = -1;
    outputThesaurusEntry(entry, outstreams.get(entry.getAlignChr()));

}
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
public SAMStats(boolean bowtie1, boolean bowtie2){
this.bowtie1 = bowtie1;
this.bowtie2 = bowtie2;
histo = new RealValuedHistogram(0, 1000, 100);
SAMFileReader reader = new SAMFileReader(System.in);
reader.setValidationStringency(ValidationStringency.SILENT);
      CloseableIterator<SAMRecord> iter = reader.iterator();
      while (iter.hasNext()) {
          SAMRecord record = iter.next();
          if(readLen==-1)
          	readLen = record.getReadLength();
          if(bowtie1)
          	processBT1SAMRecord(record);
          else if(bowtie2)
          	processBT2SAMRecord(record);
          else
          	processSAMRecord(record);
      }
      iter.close();
      reader.close();        
  }
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
protected void countReads() {
readLength=-1;
totalHits=0;
totalWeight=0;

SAMFileReader reader = new SAMFileReader(inFile);
CloseableIterator<SAMRecord> iter = reader.iterator();
Collection<SAMRecord> byRead = new ArrayList<SAMRecord>();
String lastread = null;
while (iter.hasNext()) {
    currID++;
    SAMRecord record = iter.next();
    if(readLength ==-1)
    	readLength = record.getReadLength();
    
    if (record.getReadUnmappedFlag()) {continue; }
    if (lastread == null || !lastread.equals(record.getReadName())) {
    	processRead(byRead);
    	byRead.clear();
    }
    lastread = record.getReadName();
    byRead.add(record);
	    
}
processRead(byRead);
iter.close();
reader.close();
populateArrays();
  }
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
public static SAMRecord convertToPairedEndFragment(SAMRecord rec) {
	int insertSize = Math.abs(rec.getInferredInsertSize());
	//TODO: possible to remove (insertSize > 0) clause? What does an insert size of 0 mean?
	if (rec.getReadPairedFlag() && rec.getProperPairFlag() && !rec.getReadUnmappedFlag() 
			&& (insertSize <= MAX_INSERT) && (rec.getAlignmentStart() < rec.getMateAlignmentStart())
			&& rec.getReferenceName() != "chrM") // current paired end representation doesn't do well with circular chromosomes
	{
		//We need to get the full fragment contained by read.getStart to pair.getEnd
		int readEnd=rec.getAlignmentEnd();
		int mateStart=rec.getMateAlignmentStart();
					
		//int extension = insertSize - rec.getReadLength();
		int extension=(mateStart-readEnd)+rec.getReadLength();
		if (extension <= MAX_INSERT) {
			String newRead = rec.getReadString() + StringUtils.repeat("N", extension);
			rec.setReadString(newRead);
			String newQual = StringUtils.repeat("A", newRead.length());
			if(!rec.getBaseQualityString().equals("*")) newQual = rec.getBaseQualityString() + StringUtils.repeat("A", extension);
			rec.setBaseQualityString(newQual);
			String newCigar=newRead.length()+"M";
			rec.setCigarString(newCigar);

			// Change attributes to represent single read
			rec.setMateReferenceName("*");
			rec.setMateAlignmentStart(0);
			rec.setFirstOfPairFlag(false);
			rec.setMateNegativeStrandFlag(false);
			rec.setMateUnmappedFlag(false);
			rec.setProperPairFlag(false);
			rec.setReadPairedFlag(false);
			rec.setSecondOfPairFlag(false);
		}
	} else {
		rec = null;
	}
	return rec;
}
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
private void readNext() throws SequenceFormatException {
	
	SAMRecord record;
	
	while (true) {
		
		if (!it.hasNext()) {
			nextSequence = null;
			try {
				br.close();
				fis.close();
			}
			catch (IOException ioe) {
				log.error(ioe, ioe);
			}
			return;
		}
	
		try {
			record = it.next();
		}
		catch (SAMFormatException sfe) {
			throw new SequenceFormatException(sfe.getMessage());
		}
	
		// We skip over entries with no mapping if that's what the user asked for
		if (onlyMapped && record.getReadUnmappedFlag()) {
			continue;
		}
		break;
		
	}
	
	if (recordSize == 0) {
		recordSize = (record.getReadLength()*2)+150;
		if (br.isBinary()) {
			recordSize /= 4;
		}
	}

	nextSequence = record;

}
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
/**
 * Get the reads from the appropriate source (implementation-specific).
 * Loads data to the fivePrimesList and hitsCountList
 */
public void sourceReads() {
	this.initialize();
	SAMFileReader reader = new SAMFileReader(file);
	reader.setValidationStringency(ValidationStringency.SILENT);
	CloseableIterator<SAMRecord> iter = reader.iterator();
	while (iter.hasNext()) {
	    SAMRecord record = iter.next();
	    
	    if (record.getReadUnmappedFlag()) {continue; }
	    float weight = 1/(float)record.getIntegerAttribute("NH");
	    
	    Read currRead = new Read();
        
	    List<AlignmentBlock> blocks = record.getAlignmentBlocks();
	    for(int a=0; a<blocks.size(); a++){ //Iterate over alignment blocks
	    	AlignmentBlock currBlock = blocks.get(a);
	    	int aStart = currBlock.getReferenceStart();
	    	int aEnd = aStart + currBlock.getLength()-1;
	    	int aLen = currBlock.getLength();
	    	boolean nearbyBlocks=true;
	    	while(nearbyBlocks && a<blocks.size()-1){
	    		if(blocks.get(a+1).getReferenceStart() - currBlock.getReferenceStart() < record.getReadLength()){
	    			aEnd = blocks.get(a+1).getReferenceStart() + blocks.get(a+1).getLength()-1;
	    			aLen += blocks.get(a+1).getLength();
	    			a++;
	    		}else{
	    			nearbyBlocks=false;
	    		}
	    	}
	    	
	    	ReadHit currHit = new ReadHit(
	    			record.getReferenceName().replaceFirst("^chr", ""),
	    			aStart, aEnd,
	    			record.getReadNegativeStrandFlag() ? '-' : '+',
	    			weight);
	   
	    	currRead.addHit(currHit);
		}	
	    addHits(currRead);
	}
	iter.close();
	reader.close();
   }
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
protected void countReads() {
readLength=-1;
totalHits=0;
totalWeight=0;

SAMFileReader reader = new SAMFileReader(inFile);
reader.setValidationStringency(ValidationStringency.SILENT);
CloseableIterator<SAMRecord> iter = reader.iterator();
while (iter.hasNext()) {
    currID++;
    SAMRecord record = iter.next();
    
    if (record.getReadUnmappedFlag()) {continue; }
    float weight = 1/(float)record.getIntegerAttribute("NH");
    if(readLength ==-1)
    	readLength = record.getReadLength();
    
    Read currRead = new Read((int)totalWeight);
       
    List<AlignmentBlock> blocks = record.getAlignmentBlocks();
    for(int a=0; a<blocks.size(); a++){ //Iterate over alignment blocks
    	AlignmentBlock currBlock = blocks.get(a);
    	int aStart = currBlock.getReferenceStart();
    	int aEnd = aStart + currBlock.getLength()-1;
    	int aLen = currBlock.getLength();
    	boolean nearbyBlocks=true;
    	while(nearbyBlocks && a<blocks.size()-1){
    		if(blocks.get(a+1).getReferenceStart() - currBlock.getReferenceStart() < record.getReadLength()){
    			aEnd = blocks.get(a+1).getReferenceStart() + blocks.get(a+1).getLength()-1;
    			aLen += blocks.get(a+1).getLength();
    			a++;
    		}else{
    			nearbyBlocks=false;
    		}
    	}
    	
    	ReadHit currHit = new ReadHit(gen,
			  currID,
			  record.getReferenceName().replaceFirst("^chr", ""), 
			  aStart, aEnd, 
			  record.getReadNegativeStrandFlag() ? '-' : '+',
			  weight);
   
    	currRead.addHit(currHit);
    	currID++;
	}	
    addHits(currRead);
    totalWeight++;
}
iter.close();
reader.close();
populateArrays();
  }
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
public void processSAMRecord(SAMRecord r){
	totalReads++;
	if(!r.getReadUnmappedFlag()){
		totalHits++;
		int count = 1;  //Have to figure out something for BWA when reporting multiple alignments
		if(r.getIntegerAttribute("NH")!=null)
			count = r.getIntegerAttribute("NH");
		if(count==1 && r.getMappingQuality()!=0) //Second clause for BWA
			uniquelyMapped++;
		
		weight += 1/(float)count;

		if(r.getReadPairedFlag()){
			if(r.getMateUnmappedFlag()){
				singleEnd++;
			}else{
				pairMapped++;
				if(r.getMateReferenceName().equals(r.getReferenceName())){
					pairedEndSameChr++;
				}else{
					pairedEndDiffChr++;
				}
			}
		}else{
			singleEnd++;
		}
		
		List<AlignmentBlock> blocks = r.getAlignmentBlocks();
   		if(blocks.size()>=2){
   			junctions+=blocks.size()-1;
   		}
		
		if(!r.getNotPrimaryAlignmentFlag()){
			if(!r.getReadPairedFlag() || r.getFirstOfPairFlag()){
				LHits++;
				if(r.getReadPairedFlag() && r.getProperPairFlag()){
					properPair++;
					if(!r.getReadNegativeStrandFlag() && r.getMateNegativeStrandFlag()){
						double dist = (r.getMateAlignmentStart()+r.getReadLength())-r.getAlignmentStart();
						histo.addValue(dist);
					}
				}
			}else if(r.getSecondOfPairFlag()){
				RHits++;
			}
		}else{
			notPrimary++;
		}
	}
}
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
public void processBT1SAMRecord(SAMRecord r){
	totalReads++;
	if(!r.getReadUnmappedFlag())
		if(r.getIntegerAttribute("XM")!=null){
			int xm = r.getIntegerAttribute("XM");
			if(xm!=0)
				weight++;
		}
	else{
		totalHits++;
		int count =1; //TODO: Fix this if using bowtie for multi-mapping reads
		boolean currUnique = true;
    	
		if(count==1 && currUnique){
			uniquelyMapped++;
		}
		
		weight += 1/(float)count;
		
		if(r.getReadPairedFlag()){
			if(r.getMateUnmappedFlag()){
				singleEnd++;
			}else{
				pairMapped++;
				if(r.getMateReferenceName().equals(r.getReferenceName())){
					pairedEndSameChr++;
				}else{
					pairedEndDiffChr++;
				}
			}
		}else{
			singleEnd++;
		}
		
		List<AlignmentBlock> blocks = r.getAlignmentBlocks();
   		if(blocks.size()>=2){
   			junctions+=blocks.size()-1;
   		}
   		
		if(!r.getNotPrimaryAlignmentFlag()){
			if(!r.getReadPairedFlag() || r.getFirstOfPairFlag()){
				LHits++;
				if(r.getReadPairedFlag() && r.getProperPairFlag()){
					properPair++;
					if(!r.getReadNegativeStrandFlag() && r.getMateNegativeStrandFlag()){
						double dist = (r.getMateAlignmentStart()+r.getReadLength())-r.getAlignmentStart();
						histo.addValue(dist);
					}
				}
			}else if(r.getSecondOfPairFlag()){
				RHits++;
			}
		}else{
			notPrimary++;
		}
	}
}
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
public void processBT2SAMRecord(SAMRecord r){
	totalReads++;
	if(!r.getReadUnmappedFlag()){
		totalHits++;
		int count =1; //TODO: Fix this if using bowtie2 for multi-mapping reads
		
		int primAScore = r.getIntegerAttribute("AS");
    	int secAScore=-1000000;
    	if(r.getIntegerAttribute("XS")!=null)
    		secAScore = r.getIntegerAttribute("XS");
    	boolean currUnique = primAScore > secAScore ? true : false;
    	
		if(count==1 && currUnique){
			uniquelyMapped++;
		}
		
		weight += 1/(float)count;
		
		if(r.getReadPairedFlag()){
			if(r.getMateUnmappedFlag()){
				singleEnd++;
			}else{
				pairMapped++;
				if(r.getMateReferenceName().equals(r.getReferenceName())){
					pairedEndSameChr++;
				}else{
					pairedEndDiffChr++;
				}
			}
		}else{
			singleEnd++;
		}
		
		List<AlignmentBlock> blocks = r.getAlignmentBlocks();
   		if(blocks.size()>=2){
   			junctions+=blocks.size()-1;
   		}
   		
		if(!r.getNotPrimaryAlignmentFlag()){
			if(!r.getReadPairedFlag() || r.getFirstOfPairFlag()){
				LHits++;
				if(r.getReadPairedFlag() && r.getProperPairFlag()){
					properPair++;
					if(!r.getReadNegativeStrandFlag() && r.getMateNegativeStrandFlag()){
						double dist = (r.getMateAlignmentStart()+r.getReadLength())-r.getAlignmentStart();
						histo.addValue(dist);
					}
				}
			}else if(r.getSecondOfPairFlag()){
				RHits++;
			}
		}else{
			notPrimary++;
		}
	}
}