Java SAMRecord.getReadUnmappedFlag方法代码示例

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();
	Collection<SAMRecord> byRead = new ArrayList<SAMRecord>();
	String lastread = null;
	while (iter.hasNext()) {
	    SAMRecord record = iter.next();
	    
	    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();
   }
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
private Alignment _next()
{

	SAMRecord record;
	while (iter.hasNext())
	{
	try {
		record=iter.next();
		}
		catch (SAMFormatException e)
		{
			System.err.println("catch SAMFormatException");   	
			continue;
		};
	if(record.getReadUnmappedFlag()) continue; 
	SingleEndAlignment retv=new SingleEndAlignment(record);
	return retv;

	}
	return null;
	

}
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
@Override
public void run() {

    theslog.log(true, "Pass one: " + bamfile.getAbsolutePath());

    // set up input/output objects using SAM library
    SAMFileReader inputSam = new SAMFileReader(bamfile);
    inputSam.setValidationStringency(SAMFileReader.ValidationStringency.SILENT);

    for (final SAMRecord record : inputSam) {

        if (record.getReadUnmappedFlag()) {
            // if read is unaligned, output coordinates into a bed file
            processOneUnalignedRecord(record);
        } else {
            // if aligned, then make thesaurus entries
            int mapqual = record.getMappingQuality();
            // only consider reads with a minimum mapping quality
            // (this here is not a "real" mapping quality, but a proxy for number of mismatches)
            if (mapqual >= minmapqual) {
                // check that read is aligned                
                processOneRecord(record);
            }
        }
    }

    // and close the input file stream
    inputSam.close();
}
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
@Override
public void processSequence(SAMRecord sequence) {
	
	// extract the method used for generating the SAM/BAM file if present in the header file.
	if(!headerParsed) {
		String fullHeader = sequence.getHeader().getTextHeader();
		if(fullHeader != null) {
			String[] headerLines = fullHeader.split("@");
			for(int i=0; i<headerLines.length; i++) {
				if(headerLines[i].startsWith("PG")) {
					command += headerLines[i].replace("PG\t", "").replace('\t', ' ');
				}
			}
		}
		headerParsed = true;
	}
	
	
	actualCount++;
	if (!sequence.isSecondaryOrSupplementary()) {
		++primaryCount;
	}
	
	if (sequence.getReadPairedFlag()) {
		pairedCount++;
		if (sequence.getProperPairFlag()) properPairCount++;
		if (sequence.getMateUnmappedFlag() && ! sequence.getReadUnmappedFlag()) singletonCount++;
	}
	
	if (sequence.getReadUnmappedFlag()) unmappedCount++;
	if (sequence.getReadFailsVendorQualityCheckFlag()) qcFailCount++;
	if (sequence.getDuplicateReadFlag()) duplicateCount++;
	
}
 

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 boolean readMapsToProbe(SAMRecord read, SAMFileHeader header) {
	if (read.getReadUnmappedFlag()) return false;
	
	int left, right;
	if (read.getReferenceName().contains("OriginalArray")) {
		left = ORIGINAL_ARRAY_LEFT;
		right = ORIGINAL_ARRAY_RIGHT;
	} else if (read.getReferenceName().contains("MainArray")) {
		left = MAIN_ARRAY_LEFT;
		right = MAIN_ARRAY_RIGHT;
	} else if (read.getReferenceName().contains("SimpleArray")) {
		left = SIMPLE_ARRAY_LEFT;
		right = SIMPLE_ARRAY_RIGHT;
	} else if (read.getReferenceName().contains("140105")) {
		left = ARRAY_140105;
		right = ARRAY_140105;
	} else if (read.getReferenceName().contains("Mouse")) {
		left = MRNA_ARRAY_LEFT;
		right = MRNA_ARRAY_RIGHT;
	} else {
		log.error("This reference name doesn't match: " + read.getReferenceName());
		return false;
	}
	
	left = T7BASES + left;
	right = header.getSequence(read.getReferenceIndex()).getSequenceLength() - right - T7BASES;
	
	if ((read.getAlignmentStart() <= left - OVERLAP) || read.getAlignmentEnd() >= right + OVERLAP) {
		return true;
	}
	
	return false;
}
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
private void addStartEndIndices(SAMRecord first, SAMRecord second) {
	if (!first.getReadUnmappedFlag() && !second.getReadUnmappedFlag()) {
		int start = Math.min(first.getAlignmentStart(), second.getAlignmentStart());
		int end = Math.max(first.getAlignmentEnd(), second.getAlignmentEnd());
		startIndex[start]++;
		endIndex[end]++;
	}
}
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
@Override
public boolean filterOut(SAMRecord record) {

	if (this.freezeCountersInThread.get() == null) {
		this.freezeCountersInThread.set(false);
	}

	//remove reads with two or more alignments

	if (this.freezeCountersInThread.get() == false) this.processedReadsCounter ++;
	if (record.getReadUnmappedFlag()) {

		if (this.freezeCountersInThread.get() == false) this.unmmapedReadsCounter++;
		System.err.println("UNMAPPED READ!");
		System.exit(1);
		return false;
	}

	if (onlyWithOneAlignment) {
		// bowtie 1
		if (record.getHeader().getProgramRecord("Bowtie") != null) {
			if (record.getAttribute("XM") != null
					&& ((Integer) record.getAttribute("XM")) > 1) {
				if (this.freezeCountersInThread.get() == false) this.withMoreThanOneAlignmentCounter++;
				return true;
			}
		}
		// bowtie 2
		if (record.getHeader().getProgramRecord("bowtie2") != null) {
			if (record.getAttribute("XS") != null) {
				if (this.freezeCountersInThread.get() == false) this.withMoreThanOneAlignmentCounter++;
				return true;
			}
		}
	}

	//remove ambiguous
	if (removeAmbiguous) {
		if (record.getAttribute("ZA") != null && record.getAttribute("ZA").toString().equals("Y")) {
			if (this.freezeCountersInThread.get() == false) this.ambiguousReadCounter++;
			return true;
		}
	}

	//trim to x mismatch (trim must be before bad bisulfited)
	if (trim) {
		boolean trimmed = trim(record, trimUntil);
		if (trimmed && this.freezeCountersInThread.get() == false) this.trimmedCounter++;
	}
	//bad bisulfited filter
	if (removeBad) {
		if (record.getReadGroup().getId().equals("WATSON") && watson.matcher(record.getReadString()).find()
				||
				record.getReadGroup().getId().equals("CRICK") && crick.matcher(record.getReadString()).find()) {
			if (this.freezeCountersInThread.get() == false) this.badBisulfitedCounter++;
			return true;
		}
	}
	return false;
}
 

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
@Override
public void processSequence(SAMRecord read) {
	
	if (read.getReadUnmappedFlag()) return;
	
	int leftClip = 0;
	int rightClip = 0;
	
	
	// Get the CIGAR list
	Cigar cigar = read.getCigar();
	if (cigar == null || read.getCigarLength() == 0) {
		log.debug("Read " + read.getReadString() + " does not have Cigar string.");
		return;
	}
	
	
	List<CigarElement> elements = cigar.getCigarElements();
	CigarElement first = elements.get(0);
	CigarElement last = elements.get(elements.size()-1);

	if (first.getOperator().equals(CigarOperator.S)) {
		if (read.getReadNegativeStrandFlag()) {
			rightClip = first.getLength();
		}
		else {
			leftClip = first.getLength();				
		}
	}
	
	if (last.getOperator().equals(CigarOperator.S)) {
		if (read.getReadNegativeStrandFlag()) {
			leftClip = last.getLength();
		}
		else {
			rightClip = last.getLength();
		}			
	}

	int max=leftClip;
	if (rightClip>leftClip)max=rightClip;
	
	if (max+1 > leftClipCounts.length) expandCounts(max+1);
	
	leftClipCounts[leftClip]++;
	rightClipCounts[rightClip]++;
	
}
 

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++;
		}
	}
}