/**
* Finds the adaptor boundary around the read and returns the first base inside the adaptor that
* is closest to the read boundary. If the read is in the positive strand, this is the first base
* after the end of the fragment (Picard calls it 'insert'), if the read is in the negative
* strand, this is the first base before the beginning of the fragment.
*
* <p>There are two cases we need to treat here:
*
* <p>1) Our read is in the reverse strand :
*
* <p><----------------------| * |--------------------->
*
* <p>in these cases, the adaptor boundary is at the mate start (minus one)
*
* <p>2) Our read is in the forward strand :
*
* <p>|----------------------> * <----------------------|
*
* <p>in these cases the adaptor boundary is at the start of the read plus the inferred insert
* size (plus one)
*
* @param read the read being tested for the adaptor boundary
* @return the reference coordinate for the adaptor boundary (effectively the first base IN the
* adaptor, closest to the read. NULL if the read is unmapped or the mate is mapped to another
* contig.
*/
public static Integer getAdaptorBoundary(final SAMRecord read) {
final int MAXIMUM_ADAPTOR_LENGTH = 8;
final int insertSize =
Math.abs(
read
.getInferredInsertSize()); // the inferred insert size can be negative if the mate
// is mapped before the read (so we take the absolute
// value)
if (insertSize == 0
|| read
.getReadUnmappedFlag()) // no adaptors in reads with mates in another chromosome or
// unmapped pairs
return null;
Integer
adaptorBoundary; // the reference coordinate for the adaptor boundary (effectively the first
// base IN the adaptor, closest to the read)
if (read.getReadNegativeStrandFlag())
adaptorBoundary = read.getMateAlignmentStart() - 1; // case 1 (see header)
else adaptorBoundary = read.getAlignmentStart() + insertSize + 1; // case 2 (see header)
if ((adaptorBoundary < read.getAlignmentStart() - MAXIMUM_ADAPTOR_LENGTH)
|| (adaptorBoundary > read.getAlignmentEnd() + MAXIMUM_ADAPTOR_LENGTH))
adaptorBoundary =
null; // we are being conservative by not allowing the adaptor boundary to go beyond what
// we belive is the maximum size of an adaptor
return adaptorBoundary;
}
private void collectReadData(final SAMRecord record, final ReferenceSequence ref) {
metrics.TOTAL_READS++;
readLengthHistogram.increment(record.getReadBases().length);
if (!record.getReadFailsVendorQualityCheckFlag()) {
metrics.PF_READS++;
if (isNoiseRead(record)) metrics.PF_NOISE_READS++;
if (record.getReadUnmappedFlag()) {
// If the read is unmapped see if it's adapter sequence
final byte[] readBases = record.getReadBases();
if (!(record instanceof BAMRecord)) StringUtil.toUpperCase(readBases);
if (isAdapterSequence(readBases)) {
this.adapterReads++;
}
} else if (doRefMetrics) {
metrics.PF_READS_ALIGNED++;
if (!record.getReadNegativeStrandFlag()) numPositiveStrand++;
if (record.getReadPairedFlag() && !record.getMateUnmappedFlag()) {
metrics.READS_ALIGNED_IN_PAIRS++;
// Check that both ends have mapq > minimum
final Integer mateMq = record.getIntegerAttribute("MQ");
if (mateMq == null
|| mateMq >= MAPPING_QUALITY_THRESOLD
&& record.getMappingQuality() >= MAPPING_QUALITY_THRESOLD) {
++this.chimerasDenominator;
// With both reads mapped we can see if this pair is chimeric
if (Math.abs(record.getInferredInsertSize()) > maxInsertSize
|| !record.getReferenceIndex().equals(record.getMateReferenceIndex())) {
++this.chimeras;
}
}
}
}
}
}
private static void updateSAM(
SAMRecord rec,
ReferenceSequence sequence,
SAMProgramRecord programRecord,
AlignHeapNode bestAlignHeapNode,
SRMAUtil.Space space,
String read,
String qualities,
String softClipStartBases,
String softClipStartQualities,
String softClipEndBases,
String softClipEndQualities,
boolean strand,
boolean correctBases)
throws Exception {
AlignHeapNode curAlignHeapNode = null;
AlignHeapNode prevAlignHeapNode = null;
int alignmentStart = 0;
int readIndex = -1;
byte readBases[] = null;
byte baseQualities[] = null;
byte colorErrors[] = null;
int i;
int numEdits = 0;
List<String> optFieldTags = new LinkedList<String>();
List<Object> optFieldValues = new LinkedList<Object>();
Object attr;
// Debugging stuff
String readName = rec.getReadName();
if (null == bestAlignHeapNode) {
// Do not modify the alignment
return;
}
// To generate a new CIGAR
List<CigarElement> cigarElements = null;
CigarOperator prevCigarOperator = null, curCigarOperator = null;
int prevCigarOperatorLength = 0;
// TODO
// setInferredInsertSize (invalidates paired end reads)
// setMappingQuality (?)
// setFlag
// update base qualities for color space reads
// clear attributes, but save some
Align.clearAttributes(rec, optFieldTags, optFieldValues);
readBases = new byte[read.length()];
baseQualities = new byte[qualities.length()];
for (i = 0; i < qualities.length(); i++) {
// Must subtract 33 for PHRED scaling
baseQualities[i] = (byte) (qualities.charAt(i) - 33);
}
if (strand) {
readIndex = 0;
} else {
readIndex = read.length() - 1;
}
cigarElements = new LinkedList<CigarElement>();
if (strand) { // reverse strand is the current position
alignmentStart = bestAlignHeapNode.node.position;
} else {
alignmentStart = bestAlignHeapNode.startPosition;
}
assert null != bestAlignHeapNode;
curAlignHeapNode = bestAlignHeapNode;
while (null != curAlignHeapNode) {
// Get the current cigar operator
if (null != prevAlignHeapNode
&& CigarOperator.DELETION != prevCigarOperator
&& 1 < Math.abs(curAlignHeapNode.node.position - prevAlignHeapNode.node.position)) {
curCigarOperator = CigarOperator.DELETION;
} else {
switch (curAlignHeapNode.node.type) {
case Node.MISMATCH: // Fall through
case Node.MATCH:
curCigarOperator = CigarOperator.MATCH_OR_MISMATCH;
break;
case Node.INSERTION:
// System.out.println("INS");
curCigarOperator = CigarOperator.INSERTION;
break;
default:
throw new Exception("Unknown node type");
}
if (space == SRMAUtil.Space.COLORSPACE || correctBases) {
readBases[readIndex] = (byte) curAlignHeapNode.node.base;
if (strand) {
readIndex++;
} else {
readIndex--;
}
// count the number of mismatches
switch (curAlignHeapNode.node.type) {
case Node.MISMATCH:
case Node.INSERTION:
numEdits++;
break;
default:
break;
}
} else {
// count the number of mismatches
switch (curAlignHeapNode.node.type) {
case Node.MATCH:
if (read.charAt(curAlignHeapNode.readOffset) != curAlignHeapNode.node.base) {
numEdits++;
}
break;
case Node.MISMATCH: // Fall through
if (read.charAt(curAlignHeapNode.readOffset)
!= sequence.getBases()[curAlignHeapNode.node.position - 1]) {
numEdits++;
}
break;
case Node.INSERTION:
numEdits++;
break;
default:
break;
}
}
}
if (prevCigarOperator != curCigarOperator) {
// different cigar operator
// add the previous cigar operator
if (null != prevCigarOperator) {
if (strand) { // reverse
// append
cigarElements.add(new CigarElement(prevCigarOperatorLength, prevCigarOperator));
} else {
// prepend
cigarElements.add(0, new CigarElement(prevCigarOperatorLength, prevCigarOperator));
}
}
// update prevCigarOperator
prevCigarOperator = curCigarOperator;
if (curCigarOperator == CigarOperator.DELETION) {
// length of deletion
prevCigarOperatorLength =
Math.abs(curAlignHeapNode.node.position - prevAlignHeapNode.node.position) - 1;
numEdits += prevCigarOperatorLength; // deletions
} else {
prevCigarOperatorLength = 1;
}
} else {
// same cigar operator
prevCigarOperatorLength++;
}
// Update
if (CigarOperator.DELETION != curCigarOperator) {
prevAlignHeapNode = curAlignHeapNode;
curAlignHeapNode = curAlignHeapNode.prev;
}
}
if (0 < prevCigarOperatorLength) {
if (null == prevCigarOperator || CigarOperator.DELETION == prevCigarOperator) {
throw new Exception("Ended with a null cigar operator or a deletion cigar operator");
}
if (strand) { // reverse
// append
cigarElements.add(new CigarElement(prevCigarOperatorLength, prevCigarOperator));
} else {
// prepend
cigarElements.add(0, new CigarElement(prevCigarOperatorLength, prevCigarOperator));
}
}
if (space == SRMAUtil.Space.COLORSPACE) { // color space, read bases already inferred
// Get color error string
colorErrors = new byte[read.length()];
char prevBase = SRMAUtil.COLORSPACE_ADAPTOR;
if (strand) { // reverse
for (i = 0; i < read.length(); i++) {
char nextBase = SRMAUtil.colorSpaceNextBase(prevBase, read.charAt(i));
if (nextBase == SRMAUtil.getCompliment((char) readBases[read.length() - i - 1])) {
colorErrors[i] = (byte) Alignment.GAP;
} else {
colorErrors[i] = (byte) read.charAt(i);
}
if (0 < i) {
// qualities are assumed to be always in the same direction as the color errors
baseQualities[read.length() - i] =
getColorQuality(
colorErrors[i - 1],
colorErrors[i],
(byte) (qualities.charAt(i - 1) - 33),
(byte) (qualities.charAt(i) - 33));
}
prevBase = SRMAUtil.getCompliment((char) readBases[read.length() - i - 1]);
}
// last color
baseQualities[0] = (byte) (qualities.charAt(read.length() - 1) - 33);
} else {
for (i = 0; i < read.length(); i++) {
char nextBase = SRMAUtil.colorSpaceNextBase(prevBase, read.charAt(i));
if (nextBase == readBases[i]) {
colorErrors[i] = (byte) Alignment.GAP;
} else {
colorErrors[i] = (byte) read.charAt(i);
}
if (0 < i) {
baseQualities[i - 1] =
getColorQuality(
colorErrors[i - 1],
colorErrors[i],
(byte) (qualities.charAt(i - 1) - 33),
(byte) (qualities.charAt(i) - 33));
}
prevBase = (char) readBases[i];
}
// last color
baseQualities[read.length() - 1] = (byte) (qualities.charAt(read.length() - 1) - 33);
}
} else if (correctBases) { // bases were corrected
if (strand) {
for (i = 0; i < read.length(); i++) {
if (readBases[i] == (byte) read.charAt(read.length() - i - 1)) {
baseQualities[i] = (byte) (qualities.charAt(read.length() - i - 1) - 33);
} else {
// TODO: how much to down-weight ?
baseQualities[i] =
(byte)
(SRMAUtil.QUAL2CHAR(
SRMAUtil.CHAR2QUAL(qualities.charAt(read.length() - i - 1))
- CORRECT_BASE_QUALITY_PENALTY)
- 33);
if (baseQualities[i] <= 0) {
baseQualities[i] = 1;
}
}
}
} else {
for (i = 0; i < read.length(); i++) {
if (readBases[i] == (byte) read.charAt(i)) {
baseQualities[i] = (byte) (qualities.charAt(i) - 33);
} else {
// TODO: how much to down-weight ?
baseQualities[i] =
(byte)
(SRMAUtil.QUAL2CHAR(
SRMAUtil.CHAR2QUAL(qualities.charAt(i)) - CORRECT_BASE_QUALITY_PENALTY)
- 33);
if (baseQualities[i] <= 0) {
baseQualities[i] = 1;
}
}
}
}
rec.setAttribute("XO", read);
rec.setAttribute("XQ", qualities);
} else { // bases not corrected
readBases = new byte[read.length()];
baseQualities = new byte[qualities.length()]; // qualities.length() == read.length()
if (strand) { // reverse
for (i = 0; i < read.length(); i++) {
readBases[i] = (byte) read.charAt(read.length() - i - 1);
baseQualities[i] = (byte) (qualities.charAt(read.length() - i - 1) - 33);
}
} else {
for (i = 0; i < read.length(); i++) {
readBases[i] = (byte) read.charAt(i);
baseQualities[i] = (byte) (qualities.charAt(i) - 33);
}
}
}
// Add in soft-clipping
if (null != softClipStartBases) { // prepend
cigarElements.add(0, new CigarElement(softClipStartBases.length(), CigarOperator.S));
byte tmpBases[] = new byte[readBases.length + softClipStartBases.length()];
System.arraycopy(readBases, 0, tmpBases, softClipStartBases.length(), readBases.length);
readBases = tmpBases;
for (i = 0; i < softClipStartBases.length(); i++) {
readBases[i] = (byte) softClipStartBases.charAt(i);
}
byte tmpQualities[] = new byte[baseQualities.length + softClipStartQualities.length()];
System.arraycopy(
baseQualities, 0, tmpQualities, softClipStartQualities.length(), baseQualities.length);
baseQualities = tmpQualities;
for (i = 0; i < softClipStartQualities.length(); i++) {
baseQualities[i] = (byte) softClipStartQualities.charAt(i);
}
}
if (null != softClipEndBases) { // append
cigarElements.add(new CigarElement(softClipEndBases.length(), CigarOperator.S));
byte tmpBases[] = new byte[readBases.length + softClipEndBases.length()];
System.arraycopy(readBases, 0, tmpBases, 0, readBases.length);
for (i = 0; i < softClipEndBases.length(); i++) {
tmpBases[i + readBases.length] = (byte) softClipEndBases.charAt(i);
}
readBases = tmpBases;
byte tmpQualities[] = new byte[baseQualities.length + softClipEndQualities.length()];
System.arraycopy(baseQualities, 0, tmpQualities, 0, baseQualities.length);
for (i = 0; i < softClipEndQualities.length(); i++) {
tmpQualities[i + baseQualities.length] = (byte) softClipEndQualities.charAt(i);
}
baseQualities = tmpQualities;
}
// Update SAM record
rec.setCigar(new Cigar(cigarElements));
rec.setAlignmentStart(alignmentStart);
rec.setReadBases(readBases);
rec.setBaseQualities(baseQualities);
// Reset saved attributes
Align.resetAttributes(rec, optFieldTags, optFieldValues);
// Set new attributes
if (space == SRMAUtil.Space.COLORSPACE) {
// set the XE attribute for colorError string
rec.setAttribute("XE", new String(colorErrors));
}
rec.setAttribute("AS", bestAlignHeapNode.score);
rec.setAttribute("XC", bestAlignHeapNode.alleleCoverageSum);
rec.setAttribute("PG", programRecord.getId());
rec.setAttribute("NM", numEdits);
}