Java SAMRecord.getReadString方法代码示例

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 String toFastq(SAMRecord value1) {
	String seq=value1.getReadString();
	if(value1.getReadNegativeStrandFlag()){
		seq=Sequence.reverseSequence(seq);
	}
	String name=value1.getReadName();
	String quality=value1.getBaseQualityString();
	return name+"\n"+seq+"\n"+"+\n"+quality;
}
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
public static Gene SAMFormatFullBED(SAMRecord sam){
	String chr=sam.getReferenceName();
   	int start=sam.getAlignmentStart()-1; //TODO: Do we need the -1, it is a left over from the method above.
   	Strand strand=sam.getReadNegativeStrandFlag() ? Strand.NEGATIVE : Strand.POSITIVE;
   	String cigar= sam.getCigarString(); //parse cigar and put the breaks into their component parts, in parallel keep track of the splices
   	Collection<Alignments> aligns=parseCigar(cigar, chr, start);
	Gene gene=new Gene( aligns);
	gene.setName(sam.getReadName());
	gene.setOrientation(strand);
	if(sam.getReadString() != null) {gene.setSequence(sam.getReadString()); }
   	return gene;
}
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
private boolean trim(SAMRecord record, int trimMismatches) {
	record.setAttribute("XT", "true");
	String sequenceCT = record.getReadString();

	int mismatches = (Integer) record.getAttribute("NM");

	if (mismatches >= trimMismatches) {

		boolean isReverse = record.getReadNegativeStrandFlag();


		int currentMismatches = 0;
		int trimPoint = isReverse ? sequenceCT.length() : 0;

		CigarIterator cigarIterator =
				new CigarIterator(record.getCigarString(), isReverse);

		MDTagIterator mdIterator = new MDTagIterator((String) record.getAttribute("MD"), isReverse);

		CigarOperator lastCigarOperator = null;
		while (cigarIterator.hasNext() && currentMismatches < trimMismatches) {
			CigarOperator operator = cigarIterator.next();

			switch (operator) {
				case M:
					MDTagOperator mdOperator = mdIterator.next();
					if (mdOperator == MDTagOperator.VARIANT) {
						currentMismatches++;
						if (currentMismatches < trimMismatches) {
							trimPoint = trimPoint + (isReverse ? -1 : 1);
						}
					} else if (mdOperator == MDTagOperator.SEQUENCE_MATCH) {
						// do nothing
						trimPoint = trimPoint + (isReverse ? -1 : 1);
					} else {
						//should not happen
						return false; // do not trim
					}

					break;
				case I:
					currentMismatches++;
					if (currentMismatches < trimMismatches) {
						trimPoint = trimPoint + (isReverse ? -1 : 1);
					}
					break;
				case D:
					currentMismatches++;
					mdIterator.next();
					break;
				default:
					return false; //do not trim this read, since it has non-supported CIGAR operations
			}
			lastCigarOperator = operator;
		}

		if (currentMismatches >= trimMismatches) {
			if (isReverse) {
				//trimPoint ++; //trim the last mismatch
				record.setReadString(record.getReadString().substring(0, trimPoint).replaceAll(".", "" +
						TRIMMED_BASE)
						+ record.getReadString().substring(trimPoint));
			} else {
				//trimPoint --; //trim the last mismatch
				record.setReadString(record.getReadString().substring(0, trimPoint) + record.getReadString()
						.substring
								(trimPoint).replaceAll(".", "" + TRIMMED_BASE));
			}
			return true;
		}

	}
	return false;

}
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
/**
 * Get sam line for record
 * @param rec The record
 * @return Sam formatted line ending in newline
 */
public static String getSamString(SAMRecord rec) {
	
	// Make string representation of record in sam format
	String rtrn = rec.getReadName() + "\t";
	
	// Get sam flags
	int flags = 0;
	// 0x1 template having multiple segments in sequencing
	if(rec.getReadPairedFlag()) flags += 1;
	// 0x4 segment unmapped
	if(rec.getReadUnmappedFlag()) flags += 4;
	// 0x10 SEQ being reverse complemented
	if(rec.getReadNegativeStrandFlag()) flags += 16;
	// 0x100 secondary alignment
	if(rec.getNotPrimaryAlignmentFlag()) flags += 256;
	// 0x200 not passing quality controls
	if(rec.getReadFailsVendorQualityCheckFlag()) flags += 512;
	// 0x400 PCR or optical duplicate
	if(rec.getDuplicateReadFlag()) flags += 1024;
	
	if(rec.getReadPairedFlag()) {
		// 0x2 each segment properly aligned according to the aligner
		if(rec.getProperPairFlag()) flags += 2;
		// 0x8 next segment in the template unmapped
		if(rec.getMateUnmappedFlag()) flags += 8;
		// 0x20 SEQ of the next segment in the template being reversed
		if(rec.getMateNegativeStrandFlag()) flags += 32;
		// 0x40 the first segment in the template
		if(rec.getFirstOfPairFlag()) flags += 64;
		// 0x80 the last segment in the template
		if(rec.getSecondOfPairFlag()) flags += 128;
	}
	
	rtrn += Integer.valueOf(flags).toString() + "\t";
	
	// The rest of the sam fields
	rtrn += rec.getReferenceName() + "\t";
	rtrn += rec.getAlignmentStart() + "\t";
	rtrn += rec.getMappingQuality() + "\t";
	rtrn += rec.getCigarString() + "\t";
	rtrn += rec.getMateReferenceName() + "\t";
	rtrn += rec.getMateAlignmentStart() + "\t";
	rtrn += rec.getInferredInsertSize() + "\t";
	rtrn += rec.getReadString() + "\t";
	rtrn += rec.getBaseQualityString() + "\n";
	
	return rtrn;
}
 

import net.sf.samtools.SAMRecord; //导入方法依赖的package包/类
public WrapSamRecord(SAMRecord record){
	
	this.mappingQuality=record.getMappingQuality();
	this.isDuplicate=record.getDuplicateReadFlag();
	this.readSequence=record.getReadString();
}