private GATKSAMRecord revertSoftClippedBases(GATKSAMRecord read) {
GATKSAMRecord unclipped = (GATKSAMRecord) read.clone();
Cigar unclippedCigar = new Cigar();
int matchesCount = 0;
for (CigarElement element : read.getCigar().getCigarElements()) {
if (element.getOperator() == CigarOperator.SOFT_CLIP
|| element.getOperator() == CigarOperator.MATCH_OR_MISMATCH)
matchesCount += element.getLength();
else if (matchesCount > 0) {
unclippedCigar.add(new CigarElement(matchesCount, CigarOperator.MATCH_OR_MISMATCH));
matchesCount = 0;
unclippedCigar.add(element);
} else unclippedCigar.add(element);
}
if (matchesCount > 0)
unclippedCigar.add(new CigarElement(matchesCount, CigarOperator.MATCH_OR_MISMATCH));
unclipped.setCigar(unclippedCigar);
final int newStart =
read.getAlignmentStart() + calculateAlignmentStartShift(read.getCigar(), unclippedCigar);
unclipped.setAlignmentStart(newStart);
if (newStart <= 0) {
// if the start of the unclipped read occurs before the contig,
// we must hard clip away the bases since we cannot represent reads with
// negative or 0 alignment start values in the SAMRecord (e.g., 0 means unaligned)
return hardClip(unclipped, 0, -newStart);
} else {
return unclipped;
}
}
/**
* @param read a read containing the variant
* @return the number of hard clipped and low qual bases at the read start (where start is the
* leftmost end w.r.t. the reference)
*/
public static int getNumClippedBasesAtStart(final SAMRecord read) {
// check for hard clips (never consider these bases):
final Cigar c = read.getCigar();
final CigarElement first = c.getCigarElement(0);
int numStartClippedBases = 0;
if (first.getOperator() == CigarOperator.H) {
numStartClippedBases = first.getLength();
}
final byte[] unclippedReadBases = read.getReadBases();
final byte[] unclippedReadQuals = read.getBaseQualities();
// Do a stricter base clipping than provided by CIGAR string, since this one may be too
// conservative,
// and may leave a string of Q2 bases still hanging off the reads.
// TODO: this code may not even get used because HaplotypeCaller already hard clips low quality
// tails
for (int i = numStartClippedBases; i < unclippedReadBases.length; i++) {
if (unclippedReadQuals[i] < PairHMMIndelErrorModel.BASE_QUAL_THRESHOLD)
numStartClippedBases++;
else break;
}
return numStartClippedBases;
}
/**
* Checks if a hard clipped cigar left a read starting or ending with deletions or gap (N) and
* cleans it up accordingly.
*
* @param cigar the original cigar
* @return an object with the shifts (see CigarShift class)
*/
private CigarShift cleanHardClippedCigar(final Cigar cigar) {
final Cigar cleanCigar = new Cigar();
int shiftFromStart = 0;
int shiftFromEnd = 0;
Stack<CigarElement> cigarStack = new Stack<CigarElement>();
final Stack<CigarElement> inverseCigarStack = new Stack<CigarElement>();
for (final CigarElement cigarElement : cigar.getCigarElements()) cigarStack.push(cigarElement);
for (int i = 1; i <= 2; i++) {
int shift = 0;
int totalHardClip = 0;
boolean readHasStarted = false;
boolean addedHardClips = false;
while (!cigarStack.empty()) {
CigarElement cigarElement = cigarStack.pop();
if (!readHasStarted
&& cigarElement.getOperator() != CigarOperator.DELETION
&& cigarElement.getOperator() != CigarOperator.SKIPPED_REGION
&& cigarElement.getOperator() != CigarOperator.HARD_CLIP) readHasStarted = true;
else if (!readHasStarted && cigarElement.getOperator() == CigarOperator.HARD_CLIP)
totalHardClip += cigarElement.getLength();
else if (!readHasStarted && cigarElement.getOperator() == CigarOperator.DELETION)
totalHardClip += cigarElement.getLength();
else if (!readHasStarted && cigarElement.getOperator() == CigarOperator.SKIPPED_REGION)
totalHardClip += cigarElement.getLength();
if (readHasStarted) {
if (i == 1) {
if (!addedHardClips) {
if (totalHardClip > 0)
inverseCigarStack.push(new CigarElement(totalHardClip, CigarOperator.HARD_CLIP));
addedHardClips = true;
}
inverseCigarStack.push(cigarElement);
} else {
if (!addedHardClips) {
if (totalHardClip > 0)
cleanCigar.add(new CigarElement(totalHardClip, CigarOperator.HARD_CLIP));
addedHardClips = true;
}
cleanCigar.add(cigarElement);
}
}
}
// first pass (i=1) is from end to start of the cigar elements
if (i == 1) {
shiftFromEnd = shift;
cigarStack = inverseCigarStack;
}
// second pass (i=2) is from start to end with the end already cleaned
else {
shiftFromStart = shift;
}
}
return new CigarShift(cleanCigar, shiftFromStart, shiftFromEnd);
}
/**
* Decode a single line in a SAM text file.
*
* @param line line to decode.
* @return A SAMReadFeature modeling that line.
*/
@Override
public SAMReadFeature decode(String line) {
// we may be asked to process a header line; ignore it
if (line.startsWith("@")) return null;
String[] tokens = new String[expectedTokenCount];
// split the line
int count = ParsingUtils.splitWhitespace(line, tokens);
// check to see if we've parsed the string into the right number of tokens (expectedTokenCount)
if (count != expectedTokenCount)
throw new CodecLineParsingException(
"the SAM read line didn't have the expected number of tokens "
+ "(expected = "
+ expectedTokenCount
+ ", saw = "
+ count
+ " on "
+ "line = "
+ line
+ ")");
final String readName = tokens[0];
final int flags = Integer.parseInt(tokens[1]);
final String contigName = tokens[2];
final int alignmentStart = Integer.parseInt(tokens[3]);
final int mapQ = Integer.parseInt(tokens[4]);
final String cigarString = tokens[5];
final String mateContigName = tokens[6];
final int mateAlignmentStart = Integer.parseInt(tokens[7]);
final int inferredInsertSize = Integer.parseInt(tokens[8]);
final byte[] bases = StringUtil.stringToBytes(tokens[9]);
final byte[] qualities = StringUtil.stringToBytes(tokens[10]);
// Infer the alignment end.
Cigar cigar = TextCigarCodec.decode(cigarString);
int alignmentEnd = alignmentStart + cigar.getReferenceLength() - 1;
// Remove printable character conversion from the qualities.
for (byte quality : qualities) quality -= 33;
return new SAMReadFeature(
readName,
flags,
contigName,
alignmentStart,
alignmentEnd,
mapQ,
cigarString,
mateContigName,
mateAlignmentStart,
inferredInsertSize,
bases,
qualities);
}
public static int getReadCoordinateForReferenceCoordinate(
final int alignmentStart,
final Cigar cigar,
final int refCoord,
final ClippingTail tail,
final boolean allowGoalNotReached) {
final Pair<Integer, Boolean> result =
getReadCoordinateForReferenceCoordinate(
alignmentStart, cigar, refCoord, allowGoalNotReached);
int readCoord = result.getLeft();
// Corner case one: clipping the right tail and falls on deletion, move to the next
// read coordinate. It is not a problem for the left tail because the default answer
// from getReadCoordinateForReferenceCoordinate is to give the previous read coordinate.
if (result.getRight() && tail == ClippingTail.RIGHT_TAIL) {
readCoord++;
}
// clipping the left tail and first base is insertion, go to the next read coordinate
// with the same reference coordinate. Advance to the next cigar element, or to the
// end of the read if there is no next element.
final CigarElement firstElementIsInsertion = readStartsWithInsertion(cigar);
if (readCoord == 0 && tail == ClippingTail.LEFT_TAIL && firstElementIsInsertion != null) {
readCoord = Math.min(firstElementIsInsertion.getLength(), cigar.getReadLength() - 1);
}
return readCoord;
}
/**
* Calculates the reference coordinate for the end of the read taking into account soft clips but
* not hard clips.
*
* <p>Note: getUnclippedEnd() adds soft and hard clips, this function only adds soft clips.
*
* @param read the read
* @param cigar the read's cigar
* <p>Note: this overload of the function takes the cigar as input for speed because getCigar
* is an expensive operation. Most callers should use the overload that does not take the
* cigar.
* @return the unclipped end of the read taking soft clips (but not hard clips) into account
*/
public static int getSoftEnd(final GATKRead read, final Cigar cigar) {
Utils.nonNull(read, "read");
Utils.nonNull(cigar, "cigar");
boolean foundAlignedBase = false;
int softEnd = read.getEnd();
final List<CigarElement> cigs = cigar.getCigarElements();
for (int i = cigs.size() - 1; i >= 0; --i) {
final CigarElement cig = cigs.get(i);
final CigarOperator op = cig.getOperator();
if (op
== CigarOperator
.SOFT_CLIP) { // assumes the soft clip that we found is at the end of the aligned read
softEnd += cig.getLength();
} else if (op != CigarOperator.HARD_CLIP) {
foundAlignedBase = true;
break;
}
}
if (!foundAlignedBase) { // for example 64H14S, the soft end is actually the same as the
// alignment end
softEnd = read.getEnd();
}
return softEnd;
}
/** Given a cigar string, get the number of bases hard or soft clipped at the start */
private int getNewAlignmentStartOffset(final Cigar __cigar, final Cigar __oldCigar) {
int num = 0;
for (CigarElement e : __cigar.getCigarElements()) {
if (!e.getOperator().consumesReferenceBases()) {
if (e.getOperator().consumesReadBases()) {
num += e.getLength();
}
} else {
break;
}
}
int oldNum = 0;
int curReadCounter = 0;
for (CigarElement e : __oldCigar.getCigarElements()) {
int curRefLength = e.getLength();
int curReadLength = e.getLength();
if (!e.getOperator().consumesReadBases()) {
curReadLength = 0;
}
boolean truncated = false;
if (curReadCounter + curReadLength > num) {
curReadLength = num - curReadCounter;
curRefLength = num - curReadCounter;
truncated = true;
}
if (!e.getOperator().consumesReferenceBases()) {
curRefLength = 0;
}
curReadCounter += curReadLength;
oldNum += curRefLength;
if (curReadCounter > num || truncated) {
break;
}
}
return oldNum;
}
/**
* Checks if a read starts with an insertion.
*
* @param cigarForRead the CIGAR to evaluate
* @param ignoreSoftClipOps should we ignore S operators when evaluating whether an I operator is
* at the beginning? Note that H operators are always ignored.
* @return the element if it's a leading insertion or null otherwise
*/
public static CigarElement readStartsWithInsertion(
final Cigar cigarForRead, final boolean ignoreSoftClipOps) {
for (final CigarElement cigarElement : cigarForRead.getCigarElements()) {
if (cigarElement.getOperator() == CigarOperator.INSERTION) {
return cigarElement;
} else if (cigarElement.getOperator() != CigarOperator.HARD_CLIP
&& (!ignoreSoftClipOps || cigarElement.getOperator() != CigarOperator.SOFT_CLIP)) {
break;
}
}
return null;
}
/**
* Compute the offset of the first "real" position in the cigar on the genome
*
* <p>This is defined as a first position after a run of Hs followed by a run of Ss
*
* @param cigar A non-null cigar
* @return the offset (from 0) of the first on-genome base
*/
private int calcHardSoftOffset(final Cigar cigar) {
final List<CigarElement> elements = cigar.getCigarElements();
int size = 0;
int i = 0;
while (i < elements.size() && elements.get(i).getOperator() == CigarOperator.HARD_CLIP) {
size += elements.get(i).getLength();
i++;
}
while (i < elements.size() && elements.get(i).getOperator() == CigarOperator.SOFT_CLIP) {
size += elements.get(i).getLength();
i++;
}
return size;
}
/**
* Calculates the reference coordinate for the beginning of the read taking into account soft
* clips but not hard clips.
*
* <p>Note: getUnclippedStart() adds soft and hard clips, this function only adds soft clips.
*
* @param read the read
* @param cigar the read's cigar
* <p>Note: this overload of the function takes the cigar as input for speed because getCigar
* is an expensive operation. Most callers should use the overload that does not take the
* cigar.
* @return the unclipped start of the read taking soft clips (but not hard clips) into account
*/
public static int getSoftStart(final GATKRead read, final Cigar cigar) {
Utils.nonNull(read, "read");
Utils.nonNull(cigar, "cigar");
int softStart = read.getStart();
for (final CigarElement cig : cigar.getCigarElements()) {
final CigarOperator op = cig.getOperator();
if (op == CigarOperator.SOFT_CLIP) {
softStart -= cig.getLength();
} else if (op != CigarOperator.HARD_CLIP) {
break;
}
}
return softStart;
}
@Requires({"!cigar.isEmpty()"})
private CigarShift hardClipCigar(Cigar cigar, int start, int stop) {
Cigar newCigar = new Cigar();
int index = 0;
int totalHardClipCount = stop - start + 1;
int alignmentShift = 0; // caused by hard clipping deletions
// hard clip the beginning of the cigar string
if (start == 0) {
Iterator<CigarElement> cigarElementIterator = cigar.getCigarElements().iterator();
CigarElement cigarElement = cigarElementIterator.next();
// Skip all leading hard clips
while (cigarElement.getOperator() == CigarOperator.HARD_CLIP) {
totalHardClipCount += cigarElement.getLength();
if (cigarElementIterator.hasNext()) cigarElement = cigarElementIterator.next();
else
throw new ReviewedGATKException(
"Read is entirely hardclipped, shouldn't be trying to clip it's cigar string");
}
// keep clipping until we hit stop
while (index <= stop) {
int shift = 0;
if (cigarElement.getOperator().consumesReadBases()) shift = cigarElement.getLength();
// we're still clipping or just finished perfectly
if (index + shift == stop + 1) {
alignmentShift +=
calculateHardClippingAlignmentShift(cigarElement, cigarElement.getLength());
newCigar.add(
new CigarElement(totalHardClipCount + alignmentShift, CigarOperator.HARD_CLIP));
}
// element goes beyond what we need to clip
else if (index + shift > stop + 1) {
int elementLengthAfterChopping = cigarElement.getLength() - (stop - index + 1);
alignmentShift += calculateHardClippingAlignmentShift(cigarElement, stop - index + 1);
newCigar.add(
new CigarElement(totalHardClipCount + alignmentShift, CigarOperator.HARD_CLIP));
newCigar.add(new CigarElement(elementLengthAfterChopping, cigarElement.getOperator()));
}
index += shift;
alignmentShift += calculateHardClippingAlignmentShift(cigarElement, shift);
if (index <= stop && cigarElementIterator.hasNext())
cigarElement = cigarElementIterator.next();
else break;
}
// add the remaining cigar elements
while (cigarElementIterator.hasNext()) {
cigarElement = cigarElementIterator.next();
newCigar.add(new CigarElement(cigarElement.getLength(), cigarElement.getOperator()));
}
}
// hard clip the end of the cigar string
else {
Iterator<CigarElement> cigarElementIterator = cigar.getCigarElements().iterator();
CigarElement cigarElement = cigarElementIterator.next();
// Keep marching on until we find the start
while (index < start) {
int shift = 0;
if (cigarElement.getOperator().consumesReadBases()) shift = cigarElement.getLength();
// we haven't gotten to the start yet, keep everything as is.
if (index + shift < start)
newCigar.add(new CigarElement(cigarElement.getLength(), cigarElement.getOperator()));
// element goes beyond our clip starting position
else {
int elementLengthAfterChopping = start - index;
alignmentShift +=
calculateHardClippingAlignmentShift(
cigarElement, cigarElement.getLength() - (start - index));
// if this last element is a HARD CLIP operator, just merge it with our hard clip operator
// to be added later
if (cigarElement.getOperator() == CigarOperator.HARD_CLIP)
totalHardClipCount += elementLengthAfterChopping;
// otherwise, maintain what's left of this last operator
else
newCigar.add(new CigarElement(elementLengthAfterChopping, cigarElement.getOperator()));
}
index += shift;
if (index < start && cigarElementIterator.hasNext())
cigarElement = cigarElementIterator.next();
else break;
}
// check if we are hard clipping indels
while (cigarElementIterator.hasNext()) {
cigarElement = cigarElementIterator.next();
alignmentShift +=
calculateHardClippingAlignmentShift(cigarElement, cigarElement.getLength());
// if the read had a HardClip operator in the end, combine it with the Hard Clip we are
// adding
if (cigarElement.getOperator() == CigarOperator.HARD_CLIP)
totalHardClipCount += cigarElement.getLength();
}
newCigar.add(new CigarElement(totalHardClipCount + alignmentShift, CigarOperator.HARD_CLIP));
}
return cleanHardClippedCigar(newCigar);
}
/** Given a cigar string, soft clip up to startClipEnd and soft clip starting at endClipBegin */
private Cigar softClip(final Cigar __cigar, final int __startClipEnd, final int __endClipBegin) {
if (__endClipBegin <= __startClipEnd) {
// whole thing should be soft clipped
int cigarLength = 0;
for (CigarElement e : __cigar.getCigarElements()) {
cigarLength += e.getLength();
}
Cigar newCigar = new Cigar();
newCigar.add(new CigarElement(cigarLength, CigarOperator.SOFT_CLIP));
assert newCigar.isValid(null, -1) == null;
return newCigar;
}
int curLength = 0;
Vector<CigarElement> newElements = new Vector<CigarElement>();
for (CigarElement curElem : __cigar.getCigarElements()) {
if (!curElem.getOperator().consumesReadBases()) {
if (curElem.getOperator() == CigarOperator.HARD_CLIP
|| curLength > __startClipEnd && curLength < __endClipBegin) {
newElements.add(new CigarElement(curElem.getLength(), curElem.getOperator()));
}
continue;
}
int s = curLength;
int e = curLength + curElem.getLength();
if (e <= __startClipEnd || s >= __endClipBegin) {
// must turn this entire thing into a clip
newElements.add(new CigarElement(curElem.getLength(), CigarOperator.SOFT_CLIP));
} else if (s >= __startClipEnd && e <= __endClipBegin) {
// same thing
newElements.add(new CigarElement(curElem.getLength(), curElem.getOperator()));
} else {
// we are clipping in the middle of this guy
CigarElement newStart = null;
CigarElement newMid = null;
CigarElement newEnd = null;
int midLength = curElem.getLength();
if (s < __startClipEnd) {
newStart = new CigarElement(__startClipEnd - s, CigarOperator.SOFT_CLIP);
midLength -= newStart.getLength();
}
if (e > __endClipBegin) {
newEnd = new CigarElement(e - __endClipBegin, CigarOperator.SOFT_CLIP);
midLength -= newEnd.getLength();
}
assert midLength >= 0;
if (midLength > 0) {
newMid = new CigarElement(midLength, curElem.getOperator());
}
if (newStart != null) {
newElements.add(newStart);
}
if (newMid != null) {
newElements.add(newMid);
}
if (newEnd != null) {
newElements.add(newEnd);
}
}
curLength += curElem.getLength();
}
Vector<CigarElement> finalNewElements = new Vector<CigarElement>();
CigarElement lastElement = null;
for (CigarElement elem : newElements) {
if (lastElement == null || lastElement.getOperator() != elem.getOperator()) {
if (lastElement != null) {
finalNewElements.add(lastElement);
}
lastElement = elem;
} else {
lastElement =
new CigarElement(lastElement.getLength() + elem.getLength(), lastElement.getOperator());
}
}
if (lastElement != null) {
finalNewElements.add(lastElement);
}
Cigar newCigar = new Cigar(finalNewElements);
assert newCigar.isValid(null, -1) == null;
return newCigar;
}
@Override
public String process(File page, Map<String, String> query) {
loadContigs();
if (query.get("contigName").matches("^[ACGT]+$")) {
contigs.put("manual", query.get("contigName"));
query.put("contigName", "manual");
} else if (query.get("contigName").matches("^Pf3D7.+$")) {
String[] pieces = query.get("contigName").split("[:-]");
int start = Integer.valueOf(pieces[1].replaceAll(",", ""));
int end = Integer.valueOf(pieces[2].replaceAll(",", ""));
ReferenceSequence rseq = REF.getSubsequenceAt(pieces[0], start, end);
contigs.put("manual", new String(rseq.getBases()));
query.put("contigName", "manual");
}
if (query.containsKey("contigName")
&& contigs.containsKey(query.get("contigName"))
&& graphs.containsKey(query.get("graphName"))) {
boolean showLinks = query.get("showLinks").equals("links_on");
String contig = contigs.get(query.get("contigName"));
String originalContig = contigs.get(query.get("contigName"));
String refFormattedString = "";
String kmerOrigin = "";
if (metrics.containsKey(query.get("contigName"))) {
String[] loc = metrics.get(query.get("contigName")).get("canonicalLocus").split("[:-]");
if (!loc[0].equals("*")) {
boolean isRc = metrics.get(query.get("contigName")).get("isRcCanonical").equals("1");
if (isRc) {
contig = SequenceUtils.reverseComplement(contig);
originalContig = SequenceUtils.reverseComplement(originalContig);
}
int locStart = Integer.valueOf(loc[1]);
int locEnd = Integer.valueOf(loc[2]);
Cigar cigar =
cigarStringToCigar(metrics.get(query.get("contigName")).get("cigarCanonical"));
if (cigar.getCigarElement(0).getOperator().equals(CigarOperator.S)) {
locStart -= cigar.getCigarElement(0).getLength();
}
if (cigar
.getCigarElement(cigar.getCigarElements().size() - 1)
.getOperator()
.equals(CigarOperator.S)) {
locEnd += cigar.getCigarElement(cigar.getCigarElements().size() - 1).getLength();
}
String ref = new String(REF.getSubsequenceAt(loc[0], locStart, locEnd).getBases());
StringBuilder refFormatted = new StringBuilder();
int pos = 0;
for (CigarElement ce : cigar.getCigarElements()) {
CigarOperator co = ce.getOperator();
switch (co) {
case S:
refFormatted.append(ref.substring(pos, pos + ce.getLength()).toLowerCase());
break;
case M:
refFormatted.append(ref.substring(pos, pos + ce.getLength()));
break;
case I:
refFormatted.append(StringUtils.repeat("-", ce.getLength()));
break;
}
if (ce.getOperator().consumesReferenceBases()) {
pos += ce.getLength();
}
}
refFormattedString = refFormatted.toString();
kmerOrigin = metrics.get(query.get("contigName")).get("kmerOrigin");
}
}
CortexGraph cg = graphs.get(query.get("graphName"));
String sampleName = cg.getColor(0).getSampleName();
Set<CortexLinksMap> links = new HashSet<CortexLinksMap>();
if (LINKS != null && !LINKS.isEmpty()) {
for (CortexLinksMap link : LINKS) {
if (sampleName.equals(link.getCortexLinks().getColor(0).getSampleName())) {
links.add(link);
}
}
}
Set<String> contigKmers = new HashSet<String>();
for (int i = 0; i <= contig.length() - cg.getKmerSize(); i++) {
String curKmer = contig.substring(i, i + cg.getKmerSize());
contigKmers.add(curKmer);
}
StringBuilder firstFlank = new StringBuilder();
String firstKmer = contig.substring(0, cg.getKmerSize());
Set<String> pks = CortexUtils.getPrevKmers(cg, firstKmer, 0);
Set<String> usedPrevKmers = new HashSet<String>();
usedPrevKmers.add(firstKmer);
while (pks.size() == 1 && usedPrevKmers.size() <= 100) {
String kmer = pks.iterator().next();
firstFlank.insert(0, kmer.charAt(0));
if (usedPrevKmers.contains(kmer)) {
break;
}
usedPrevKmers.add(kmer);
pks = CortexUtils.getPrevKmers(cg, kmer, 0);
}
StringBuilder lastFlank = new StringBuilder();
String lastKmer = contig.substring(contig.length() - cg.getKmerSize(), contig.length());
Set<String> nks = CortexUtils.getNextKmers(cg, lastKmer, 0);
Set<String> usedNextKmers = new HashSet<String>();
usedNextKmers.add(lastKmer);
while (nks.size() == 1 && usedNextKmers.size() <= 100) {
String kmer = nks.iterator().next();
lastFlank.append(kmer.charAt(kmer.length() - 1));
if (usedNextKmers.contains(kmer)) {
break;
}
usedNextKmers.add(kmer);
nks = CortexUtils.getNextKmers(cg, kmer, 0);
}
contig = firstFlank.toString() + contig + lastFlank.toString();
DirectedGraph<CtxVertex, MultiEdge> g =
new DefaultDirectedGraph<CtxVertex, MultiEdge>(MultiEdge.class);
for (int i = 0; i <= contig.length() - cg.getKmerSize(); i++) {
String curKmer = contig.substring(i, i + cg.getKmerSize());
CortexKmer ck = new CortexKmer(curKmer);
CtxVertex curVer =
new CtxVertex(
curKmer,
i,
contigKmers.contains(curKmer) ? VertexType.CONTIG : VertexType.CLIPPED,
cg.findRecord(ck));
g.addVertex(curVer);
String expectedPrevKmer =
(i > 0) ? contig.substring(i - 1, i - 1 + cg.getKmerSize()) : "";
String expectedNextKmer =
(i < contig.length() - cg.getKmerSize())
? contig.substring(i + 1, i + 1 + cg.getKmerSize())
: "";
Set<String> prevKmers = CortexUtils.getPrevKmers(cg, curKmer, 0);
for (String prevKmer : prevKmers) {
if (!expectedPrevKmer.equals(prevKmer)) {
CortexKmer pk = new CortexKmer(prevKmer);
CtxVertex prevVer = new CtxVertex(prevKmer, i - 1, VertexType.IN, cg.findRecord(pk));
MultiEdge me =
g.containsEdge(prevVer, curVer) ? g.getEdge(prevVer, curVer) : new MultiEdge();
me.addGraphName(cg.getCortexFile().getName());
g.addVertex(prevVer);
g.addEdge(prevVer, curVer, me);
}
}
Set<String> nextKmers = CortexUtils.getNextKmers(cg, curKmer, 0);
for (String nextKmer : nextKmers) {
if (!expectedNextKmer.equals(nextKmer)) {
CortexKmer nk = new CortexKmer(nextKmer);
CtxVertex nextVer = new CtxVertex(nextKmer, i + 1, VertexType.OUT, cg.findRecord(nk));
MultiEdge me =
g.containsEdge(curVer, nextVer) ? g.getEdge(curVer, nextVer) : new MultiEdge();
me.addGraphName(cg.getCortexFile().getName());
g.addVertex(nextVer);
g.addEdge(curVer, nextVer, me);
}
}
}
Set<Map<String, Object>> verticesWithLinks = new HashSet<Map<String, Object>>();
DataFrame<String, String, Integer> hv = new DataFrame<String, String, Integer>(0);
for (int q = 0; q <= contig.length() - cg.getKmerSize(); q++) {
// String sk = cv.getBinaryKmer();
String sk = contig.substring(q, q + cg.getKmerSize());
CortexKmer ck = new CortexKmer(sk);
for (CortexLinksMap link : links) {
if (link.containsKey(ck)) {
CortexLinksRecord clr = link.get(ck);
Map<String, Integer> lc =
(!showLinks)
? new HashMap<String, Integer>()
: CortexUtils.getKmersAndCoverageInLink(cg, sk, clr);
Map<String, Object> entry = new HashMap<String, Object>();
entry.put("kmer", sk);
entry.put("lc", lc);
verticesWithLinks.add(entry);
if (showLinks) {
for (CortexJunctionsRecord cjr : clr.getJunctions()) {
List<String> lk = CortexUtils.getKmersInLink(cg, sk, cjr);
for (int i = 0; i < lk.size(); i++) {
String kili = lk.get(i);
for (int j = 0; j < lk.size(); j++) {
String kilj = lk.get(j);
if (i != j) {
hv.set(kili, kilj, hv.get(kili, kilj) + cjr.getCoverage(0));
}
}
}
}
}
}
}
}
/*
int hvMax = 0;
Map<String, Integer> hvlin = new HashMap<String, Integer>();
if (showLinks) {
for (String kili : hv.getRowNames()) {
for (String kilj : hv.getColNames()) {
int cov = hv.get(kili, kilj);
String id = kili + "_" + kilj;
hvlin.put(id, cov);
if (cov > hvMax) {
hvMax = cov;
}
}
}
}
*/
JSONObject jo = new JSONObject();
jo.put("contig", contig);
jo.put("originalContig", originalContig);
jo.put("ref", refFormattedString);
jo.put("kmerOrigin", kmerOrigin);
jo.put("kmerSize", cg.getKmerSize());
jo.put("clipStart", firstFlank.length());
jo.put("clipEnd", contig.length() - lastFlank.length());
List<Map<String, Object>> va = new ArrayList<Map<String, Object>>();
for (CtxVertex v : g.vertexSet()) {
Map<String, Object> vm = new HashMap<String, Object>();
vm.put("base", v.getBase());
vm.put("kmer", v.getKmer());
vm.put("pos", v.getPos());
vm.put("type", v.getVertexType().name());
vm.put("missing", v.isMissingFromGraph());
vm.put("cov", v.getCoverage());
va.add(vm);
}
jo.put("vertices", va);
jo.put("verticesWithLinks", verticesWithLinks);
// jo.put("hvlin", hvlin);
// jo.put("hvmax", hvMax);
return jo.toString();
}
return null;
}
private static Pair<Integer, Boolean> getReadCoordinateForReferenceCoordinate(
final int alignmentStart,
final Cigar cigar,
final int refCoord,
final boolean allowGoalNotReached) {
int readBases = 0;
int refBases = 0;
boolean fallsInsideDeletionOrSkippedRegion = false;
boolean endJustBeforeDeletionOrSkippedRegion = false;
boolean fallsInsideOrJustBeforeDeletionOrSkippedRegion = false;
final int goal = refCoord - alignmentStart; // The goal is to move this many reference bases
if (goal < 0) {
if (allowGoalNotReached) {
return new MutablePair<>(CLIPPING_GOAL_NOT_REACHED, false);
} else {
throw new GATKException(
"Somehow the requested coordinate is not covered by the read. Too many deletions?");
}
}
boolean goalReached = refBases == goal;
final Iterator<CigarElement> cigarElementIterator = cigar.getCigarElements().iterator();
while (!goalReached && cigarElementIterator.hasNext()) {
final CigarElement cigarElement = cigarElementIterator.next();
int shift = 0;
if (cigarElement.getOperator().consumesReferenceBases()
|| cigarElement.getOperator() == CigarOperator.SOFT_CLIP) {
if (refBases + cigarElement.getLength() < goal) {
shift = cigarElement.getLength();
} else {
shift = goal - refBases;
}
refBases += shift;
}
goalReached = refBases == goal;
if (!goalReached && cigarElement.getOperator().consumesReadBases()) {
readBases += cigarElement.getLength();
}
if (goalReached) {
// Is this base's reference position within this cigar element? Or did we use it all?
final boolean endsWithinCigar = shift < cigarElement.getLength();
// If it isn't, we need to check the next one. There should *ALWAYS* be a next one
// since we checked if the goal coordinate is within the read length, so this is just a
// sanity check.
if (!endsWithinCigar && !cigarElementIterator.hasNext()) {
if (allowGoalNotReached) {
return new MutablePair<>(CLIPPING_GOAL_NOT_REACHED, false);
} else {
throw new GATKException(
String.format(
"Reference coordinate corresponds to a non-existent base in the read. This should never happen -- check read with alignment start: %s and cigar: %s",
alignmentStart, cigar));
}
}
CigarElement nextCigarElement = null;
// if we end inside the current cigar element, we just have to check if it is a deletion (or
// skipped region)
if (endsWithinCigar) {
fallsInsideDeletionOrSkippedRegion =
(cigarElement.getOperator() == CigarOperator.DELETION
|| cigarElement.getOperator() == CigarOperator.SKIPPED_REGION);
} // if we end outside the current cigar element, we need to check if the next element is an
// insertion, deletion or skipped region.
else {
nextCigarElement = cigarElementIterator.next();
// if it's an insertion, we need to clip the whole insertion before looking at the next
// element
if (nextCigarElement.getOperator() == CigarOperator.INSERTION) {
readBases += nextCigarElement.getLength();
if (!cigarElementIterator.hasNext()) {
if (allowGoalNotReached) {
return new MutablePair<>(CLIPPING_GOAL_NOT_REACHED, false);
} else {
throw new GATKException(
String.format(
"Reference coordinate corresponds to a non-existent base in the read. This should never happen -- check read with alignment start: %s and cigar: %s",
alignmentStart, cigar));
}
}
nextCigarElement = cigarElementIterator.next();
}
// if it's a deletion (or skipped region), we will pass the information on to be handled
// downstream.
endJustBeforeDeletionOrSkippedRegion =
(nextCigarElement.getOperator() == CigarOperator.DELETION
|| nextCigarElement.getOperator() == CigarOperator.SKIPPED_REGION);
}
fallsInsideOrJustBeforeDeletionOrSkippedRegion =
endJustBeforeDeletionOrSkippedRegion || fallsInsideDeletionOrSkippedRegion;
// If we reached our goal outside a deletion (or skipped region), add the shift
if (!fallsInsideOrJustBeforeDeletionOrSkippedRegion
&& cigarElement.getOperator().consumesReadBases()) {
readBases += shift;
} // If we reached our goal just before a deletion (or skipped region) we need
// to add the shift of the current cigar element but go back to it's last element to return
// the last
// base before the deletion (or skipped region) (see warning in function contracts)
else if (endJustBeforeDeletionOrSkippedRegion
&& cigarElement.getOperator().consumesReadBases()) {
readBases += shift - 1;
} // If we reached our goal inside a deletion (or skipped region), or just between a
// deletion and a skipped region,
// then we must backtrack to the last base before the deletion (or skipped region)
else if (fallsInsideDeletionOrSkippedRegion
|| (endJustBeforeDeletionOrSkippedRegion
&& nextCigarElement.getOperator().equals(CigarOperator.N))
|| (endJustBeforeDeletionOrSkippedRegion
&& nextCigarElement.getOperator().equals(CigarOperator.D))) {
readBases--;
}
}
}
if (!goalReached) {
if (allowGoalNotReached) {
return new MutablePair<>(CLIPPING_GOAL_NOT_REACHED, false);
} else {
throw new GATKException(
"Somehow the requested coordinate is not covered by the read. Alignment "
+ alignmentStart
+ " | "
+ cigar);
}
}
return Pair.of(readBases, fallsInsideOrJustBeforeDeletionOrSkippedRegion);
}
public void printAlignment(final byte[] ref, final byte[] read, final int width) {
final StringBuilder bread = new StringBuilder();
final StringBuilder bref = new StringBuilder();
final StringBuilder match = new StringBuilder();
int i = 0;
int j = 0;
final int offset = getAlignmentStart2wrt1();
Cigar cigar = getCigar();
if (overhangStrategy != OverhangStrategy.SOFTCLIP) {
// we need to go through all the hassle below only if we do not do softclipping;
// otherwise offset is never negative
if (offset < 0) {
for (; j < (-offset); j++) {
bread.append((char) read[j]);
bref.append(' ');
match.append(' ');
}
// at negative offsets, our cigar's first element carries overhanging bases
// that we have just printed above. Tweak the first element to
// exclude those bases. Here we create a new list of cigar elements, so the original
// list/original cigar are unchanged (they are unmodifiable anyway!)
final List<CigarElement> tweaked = new ArrayList<>();
tweaked.addAll(cigar.getCigarElements());
tweaked.set(
0,
new CigarElement(
cigar.getCigarElement(0).getLength() + offset,
cigar.getCigarElement(0).getOperator()));
cigar = new Cigar(tweaked);
}
}
if (offset
> 0) { // note: the way this implementation works, cigar will ever start from S *only* if
// read starts before the ref, i.e. offset = 0
for (; i < getAlignmentStart2wrt1(); i++) {
bref.append((char) ref[i]);
bread.append(' ');
match.append(' ');
}
}
for (final CigarElement e : cigar.getCigarElements()) {
switch (e.getOperator()) {
case M:
for (int z = 0; z < e.getLength(); z++, i++, j++) {
bref.append((i < ref.length) ? (char) ref[i] : ' ');
bread.append((j < read.length) ? (char) read[j] : ' ');
match.append(
(i < ref.length && j < read.length) ? (ref[i] == read[j] ? '.' : '*') : ' ');
}
break;
case I:
for (int z = 0; z < e.getLength(); z++, j++) {
bref.append('-');
bread.append((char) read[j]);
match.append('I');
}
break;
case S:
for (int z = 0; z < e.getLength(); z++, j++) {
bref.append(' ');
bread.append((char) read[j]);
match.append('S');
}
break;
case D:
for (int z = 0; z < e.getLength(); z++, i++) {
bref.append((char) ref[i]);
bread.append('-');
match.append('D');
}
break;
default:
throw new GATKException("Unexpected Cigar element:" + e.getOperator());
}
}
for (; i < ref.length; i++) bref.append((char) ref[i]);
for (; j < read.length; j++) bread.append((char) read[j]);
int pos = 0;
final int maxlength = Math.max(match.length(), Math.max(bread.length(), bref.length()));
while (pos < maxlength) {
print_cautiously(match, pos, width);
print_cautiously(bread, pos, width);
print_cautiously(bref, pos, width);
System.out.println();
pos += width;
}
}