/**
* Check to ensure that the alignment makes sense based on the contents of the header.
*
* @param header The SAM file header.
* @param read The read to verify.
* @return true if alignment agrees with header, false otherwise.
*/
public static boolean alignmentAgreesWithHeader(final SAMFileHeader header, final GATKRead read) {
final int referenceIndex = getReferenceIndex(read, header);
// Read is aligned to nonexistent contig
if (!read.isUnmapped() && referenceIndex == SAMRecord.NO_ALIGNMENT_REFERENCE_INDEX) {
return false;
}
final SAMSequenceRecord contigHeader = header.getSequence(referenceIndex);
// Read is aligned to a point after the end of the contig
return read.isUnmapped() || read.getStart() <= contigHeader.getSequenceLength();
}
/**
* 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. CANNOT_COMPUTE_ADAPTOR_BOUNDARY if the read is unmapped or
* the mate is mapped to another contig.
*/
public static int getAdaptorBoundary(final GATKRead read) {
if (!hasWellDefinedFragmentSize(read)) {
return CANNOT_COMPUTE_ADAPTOR_BOUNDARY;
} else if (read.isReverseStrand()) {
return read.getMateStart() - 1; // case 1 (see header)
} else {
final int insertSize =
Math.abs(
read
.getFragmentLength()); // the inferred insert size can be negative if the mate is
// mapped before the read (so we take the absolute value)
return read.getStart() + insertSize + 1; // case 2 (see header)
}
}
/**
* 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;
}
/**
* Can the adaptor sequence of read be reliably removed from the read based on the alignment of
* read and its mate?
*
* @param read the read to check
* @return true if it can, false otherwise
*/
public static boolean hasWellDefinedFragmentSize(final GATKRead read) {
if (read.getFragmentLength() == 0)
// no adaptors in reads with mates in another chromosome or unmapped pairs
{
return false;
}
if (!read.isPaired())
// only reads that are paired can be adaptor trimmed
{
return false;
}
if (read.isUnmapped() || read.mateIsUnmapped())
// only reads when both reads are mapped can be trimmed
{
return false;
}
// if ( ! read.isProperlyPaired() )
// // note this flag isn't always set properly in BAMs, can will stop us from
// eliminating some proper pairs
// // reads that aren't part of a proper pair (i.e., have strange alignments) can't
// be trimmed
// return false;
if (read.isReverseStrand() == read.mateIsReverseStrand())
// sanity check on isProperlyPaired to ensure that read1 and read2 aren't on the same strand
{
return false;
}
if (read.isReverseStrand()) {
// we're on the negative strand, so our read runs right to left
return read.getEnd() > read.getMateStart();
} else {
// we're on the positive strand, so our mate should be to our right (his start + insert size
// should be past our start)
return read.getStart() <= read.getMateStart() + read.getFragmentLength();
}
}
/**
* Is a base inside a read?
*
* @param read the read to evaluate
* @param referenceCoordinate the reference coordinate of the base to test
* @return true if it is inside the read, false otherwise.
*/
public static boolean isInsideRead(final GATKRead read, final int referenceCoordinate) {
return referenceCoordinate >= read.getStart() && referenceCoordinate <= read.getEnd();
}
