public static int getReadCoordinateForReferenceCoordinate(
final int alignmentStart,
final Cigar cigar,
final int refCoord,
final ClippingTail tail,
final boolean allowGoalNotReached) {
Pair<Integer, Boolean> result =
getReadCoordinateForReferenceCoordinate(
alignmentStart, cigar, refCoord, allowGoalNotReached);
int readCoord = result.getFirst();
// 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.getSecond() && 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.
Pair<Boolean, CigarElement> firstElementIsInsertion = readStartsWithInsertion(cigar);
if (readCoord == 0 && tail == ClippingTail.LEFT_TAIL && firstElementIsInsertion.getFirst())
readCoord =
Math.min(firstElementIsInsertion.getSecond().getLength(), cigar.getReadLength() - 1);
return readCoord;
}
public static Pair<Boolean, CigarElement> readStartsWithInsertion(final Cigar cigar) {
for (CigarElement cigarElement : cigar.getCigarElements()) {
if (cigarElement.getOperator() == CigarOperator.INSERTION)
return new Pair<Boolean, CigarElement>(true, cigarElement);
else if (cigarElement.getOperator() != CigarOperator.HARD_CLIP
&& cigarElement.getOperator() != CigarOperator.SOFT_CLIP) break;
}
return new Pair<Boolean, CigarElement>(false, null);
}
private static boolean isAfterOp(
final Cigar cigar,
final int currentOperatorIndex,
final CigarOperator op,
final boolean isFirstOp,
final boolean isFirstBaseOfOp) {
return !isFirstOp
&& isFirstBaseOfOp
&& cigar.getCigarElement(currentOperatorIndex - 1).getOperator() == op;
}
public static void main(String[] args) {
args =
new String[] {
"/commun/data/users/cfaucheron/aln_20120329/S0529/data_S0529/S0529_sort.nodup.bam"
};
ReferenceSequenceFile rsf =
ReferenceSequenceFileFactory.getReferenceSequenceFile(
new File("/commun/data/pubdb/ucsc/hg19/chromosomes/hg19.fa"));
int count = 0;
for (String filename : args) {
File file = new File(filename);
SAMFileReader samIn = new SAMFileReader(file);
SAMRecordIterator r = samIn.iterator();
while (r.hasNext()) {
SAMRecord rec = r.next();
if (rec.getReadUnmappedFlag()) continue;
if (++count > 10000) break;
if (rec.getAlignmentStart() > rec.getAlignmentEnd()) throw new IllegalStateException();
byte bases[] =
rsf.getSubsequenceAt(
rec.getReferenceName(),
rec.getAlignmentStart(),
Math.max(
rec.getAlignmentEnd(),
rec.getAlignmentStart() + rec.getCigar().getPaddedReferenceLength()))
.getBases();
Iterator<CigarAlignment> i = CigarAlignment.iterator(rec);
/*System.err.println(rec.getCigarString());
System.err.println(bases.length);
System.err.println("start:"+rec.getAlignmentStart());*/
StringBuilder s1 = new StringBuilder();
StringBuilder s2 = new StringBuilder();
while (i.hasNext()) {
CigarAlignment caln = i.next();
/*
System.err.println(rec.getCigarString());
System.err.println("bases.length:"+bases.length);
System.err.println("refpos:"+caln.getReferencePosition1());
System.err.println("readpos:"+rec.getAlignmentStart());
*/
if (caln.getReferencePosition1() - rec.getAlignmentStart() >= bases.length) {
System.out.println("SHORT!");
System.out.println("op:" + caln.getCigarOperator());
System.out.println("read start:" + rec.getAlignmentStart());
System.out.println("clan.pos1:" + caln.getReferencePosition1());
System.out.println("read end:" + rec.getAlignmentEnd());
System.out.println("bases.length:" + bases.length);
System.out.println(
"getPaddedReferenceLength:" + rec.getCigar().getPaddedReferenceLength());
System.out.println("getReferenceLength:" + rec.getCigar().getReferenceLength());
System.out.println("getReadLength:" + rec.getCigar().getReadLength());
System.out.println(
"cigar.read.length:" + Cigar.getReadLength(rec.getCigar().getCigarElements()));
count = 2000;
break;
}
if (caln.isInsertRef()) {
s2.append("-");
s1.append(caln.getReadBase());
} else if (caln.isDeletionRef()) {
s2.append((char) bases[caln.getReferencePosition1() - rec.getAlignmentStart()]);
s1.append("-");
} else {
s2.append((char) bases[caln.getReferencePosition1() - rec.getAlignmentStart()]);
s1.append(caln.getReadBase());
}
// System.out.println(s1);
// System.out.println(s2);
// System.out.println();
}
System.out.println(
rec.getCigarString() + " " + rec.getReferenceName() + ":" + rec.getAlignmentStart());
System.out.println("ref :" + new String(bases));
System.out.println("read:" + new String(rec.getReadBases()));
System.out.println();
System.out.println(s1);
System.out.println(s2);
System.out.println();
}
samIn.close();
}
}
public static Pair<Integer, Boolean> getReadCoordinateForReferenceCoordinate(
final int alignmentStart,
final Cigar cigar,
final int refCoord,
final boolean allowGoalNotReached) {
int readBases = 0;
int refBases = 0;
boolean fallsInsideDeletion = false;
int goal = refCoord - alignmentStart; // The goal is to move this many reference bases
if (goal < 0) {
if (allowGoalNotReached) {
return new Pair<Integer, Boolean>(CLIPPING_GOAL_NOT_REACHED, false);
} else {
throw new ReviewedStingException(
"Somehow the requested coordinate is not covered by the read. Too many deletions?");
}
}
boolean goalReached = refBases == goal;
Iterator<CigarElement> cigarElementIterator = cigar.getCigarElements().iterator();
while (!goalReached && cigarElementIterator.hasNext()) {
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?
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 Pair<Integer, Boolean>(CLIPPING_GOAL_NOT_REACHED, false);
} else {
throw new ReviewedStingException(
"Reference coordinate corresponds to a non-existent base in the read. This should never happen -- call Mauricio");
}
}
CigarElement nextCigarElement;
// if we end inside the current cigar element, we just have to check if it is a deletion
if (endsWithinCigar)
fallsInsideDeletion = cigarElement.getOperator() == CigarOperator.DELETION;
// if we end outside the current cigar element, we need to check if the next element is an
// insertion or deletion.
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 Pair<Integer, Boolean>(CLIPPING_GOAL_NOT_REACHED, false);
} else {
throw new ReviewedStingException(
"Reference coordinate corresponds to a non-existent base in the read. This should never happen -- call Mauricio");
}
}
nextCigarElement = cigarElementIterator.next();
}
// if it's a deletion, we will pass the information on to be handled downstream.
fallsInsideDeletion = nextCigarElement.getOperator() == CigarOperator.DELETION;
}
// If we reached our goal outside a deletion, add the shift
if (!fallsInsideDeletion && cigarElement.getOperator().consumesReadBases())
readBases += shift;
// If we reached our goal inside a deletion, but the deletion is the next cigar element then
// 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 (see warning in function contracts)
else if (fallsInsideDeletion && !endsWithinCigar) readBases += shift - 1;
// If we reached our goal inside a deletion then we must backtrack to the last base before
// the deletion
else if (fallsInsideDeletion && endsWithinCigar) readBases--;
}
}
if (!goalReached) {
if (allowGoalNotReached) {
return new Pair<Integer, Boolean>(CLIPPING_GOAL_NOT_REACHED, false);
} else {
throw new ReviewedStingException(
"Somehow the requested coordinate is not covered by the read. Alignment "
+ alignmentStart
+ " | "
+ cigar);
}
}
return new Pair<Integer, Boolean>(readBases, fallsInsideDeletion);
}