public boolean processString() {
String value = "";
SequenceReader r = new SequenceReader(true);
r.indexFASTAFile(hitFilename, null, true);
int l = (3 * querySeq.length()) / 2;
String hitSeq = r.getSubSequence(hitReference.getId(), hitStart, hitStart + l);
int hitPtr = 0;
int queryPtr = 0;
boolean displayResult = false;
boolean donePreClipping = false;
int tagCtr = 0;
int i = 0;
boolean continueParsing = true;
int totalCount = 0;
int delCount = 0;
int insCount = 0;
int matchCount = 0;
boolean processed = true;
// System.out.println(" Hit: "+hitSeq.length()+" "+hitSeq);
// System.out.println("Query: "+querySeq.length()+" "+querySeq);
hitAlnSize = 0;
queryAlnSize = 0;
hitAlnSize = 0;
while ((i < cigarString.length()) && (continueParsing)) {
// for (int i=0; i<cigarString.length(); i++) {
// System.out.println("hitPtr="+hitPtr+" queryPtr="+queryPtr);
// System.out.println("Query: " + queryString.toString());
// System.out.println(" Hit: " + hitString.toString());
char c = cigarString.charAt(i);
if (Character.isDigit(c)) {
value = value + c;
} else {
int n = Integer.parseInt(value);
totalCount += n;
// System.out.println(n + " " + c);
switch (c) {
case 'M':
case '=':
case 'X':
queryString.append(querySeq.substring(queryPtr, queryPtr + n));
hitString.append(hitSeq.substring(hitPtr, hitPtr + n));
queryPtr += n;
hitPtr += n;
queryAlnSize += n;
hitAlnSize += n;
donePreClipping = true;
matchCount += n;
break;
case 'I':
if (n > 100) {
System.out.println("");
System.out.println("Error: large I (" + n + ") - read " + queryID + " ignored");
processed = false;
continueParsing = false;
} else {
queryString.append(querySeq.substring(queryPtr, queryPtr + n));
for (int j = 0; j < n; j++) {
hitString.append('-');
}
queryPtr += n;
queryAlnSize += n;
}
donePreClipping = true;
insCount += n;
break;
case 'D':
if (n > 100) {
System.out.println("Error: large D (" + n + ") - read " + queryID + " ignored");
processed = false;
continueParsing = false;
} else {
hitString.append(hitSeq.substring(hitPtr, hitPtr + n));
for (int j = 0; j < n; j++) {
queryString.append('-');
}
hitPtr += n;
hitAlnSize += n;
}
donePreClipping = true;
delCount += n;
break;
case 'N':
System.out.println("Warning: encountered N in CIGAR format!");
System.out.println("");
displayResult = true;
hitString.append(hitSeq.substring(hitPtr, hitPtr + n));
for (int j = 0; j < n; j++) {
queryString.append('-');
}
queryPtr += n;
hitPtr += n;
donePreClipping = true;
break;
case 'S':
// System.out.println("Warnning: encountered S in CIGAR format!");
queryPtr += n;
if (!donePreClipping) {
queryStart += n;
}
displayResult = true;
break;
case 'H':
// System.out.println("Warning: encountered H in CIGAR format!");
if (!donePreClipping) {
queryStart += n;
} else {
// System.out.println("Warning: hard clipping at end");
}
displayResult = true;
break;
case 'P':
System.out.println("Warning: encountered P in CIGAR format!");
System.out.println("");
displayResult = true;
donePreClipping = true;
break;
default:
System.out.println("Unrecognised character in CIGAR string: " + c);
processed = false;
break;
}
value = "";
tagCtr++;
// System.out.println("qseq="+querySeq.length()+" matchCount="+matchCount+"
// insCount="+insCount+" delCount="+delCount+" totalCount="+totalCount);
// System.out.println("Query: "+queryString.toString());
// System.out.println(" Hit: "+hitString.toString());
}
i++;
// System.out.println("i="+i+" and length="+cigarString.length());
}
// if (displayResult) {
// System.out.println(queryFilename);
// System.out.println("Query: " + queryString.toString());
// System.out.println(" Hit: " + hitString.toString());
// System.exit(1);
// }
return processed;
}
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);
}