@Override
public boolean nextKeyValue() {
long virtPos;
while ((virtPos = bci.getFilePointer()) < virtualEnd
|| (keepReadPairsTogether && readPair && !lastOfPair)) {
final SAMRecord r = codec.decode();
if (r == null) return false;
// Since we're reading from a BAMRecordCodec directly we have to set the
// validation stringency ourselves.
if (this.stringency != null) r.setValidationStringency(this.stringency);
readPair = r.getReadPairedFlag();
if (readPair) {
boolean first = r.getFirstOfPairFlag(), second = r.getSecondOfPairFlag();
// According to the SAM spec (section 1.4) it is possible for pairs to have
// multiple segments (i.e. more than two), in which case both `first` and
// `second` will be true.
boolean firstOfPair = first && !second;
lastOfPair = !first && second;
// ignore any template that is not first in a pair right at the start of a split
// since it will have been returned in the previous split
if (virtPos == virtualStart && keepReadPairsTogether && !firstOfPair) {
continue;
}
}
if (!overlaps(r)) {
continue;
}
key.set(getKey(r));
record.set(r);
return true;
}
return false;
}
@Test
public void testWithIndividualReadBarcodes() {
final AbstractMarkDuplicatesCommandLineProgramTester tester = getTester();
final String readNameOne = "RUNID:1:1:15993:13361";
final String readNameTwo = "RUNID:2:2:15993:13362";
final String readNameThree = "RUNID:3:3:15993:13362";
// first two reads have the same barcode (all three), third read has a different barcode for the
// second end
tester.addMatePair(
readNameOne,
2,
41212324,
41212310,
false,
false,
false,
false,
"33S35M",
"19S49M",
true,
true,
false,
false,
false,
DEFAULT_BASE_QUALITY);
tester.addMatePair(
readNameTwo,
2,
41212324,
41212310,
false,
false,
true,
true,
"33S35M",
"19S49M",
true,
true,
false,
false,
false,
DEFAULT_BASE_QUALITY); // same barcode as the first
tester.addMatePair(
readNameThree,
2,
41212324,
41212310,
false,
false,
false,
false,
"33S35M",
"19S49M",
true,
true,
false,
false,
false,
DEFAULT_BASE_QUALITY);
final String barcodeTag = "BC";
final String readOneBarcodeTag =
"BX"; // want the same tag as the second end, since this is allowed
final String readTwoBarcodeTag = "BX";
for (final SAMRecord record : new IterableAdapter<SAMRecord>(tester.getRecordIterator())) {
record.setAttribute(barcodeTag, "Barcode1"); // same barcode
if (record.getFirstOfPairFlag()) { // always the same value for the first end
record.setAttribute(readOneBarcodeTag, "readOne1");
} else { // second end
if (record.getReadName().equals(readNameOne) || record.getReadName().equals(readNameTwo)) {
record.setAttribute(readTwoBarcodeTag, "readTwo1");
} else if (record.getReadName().equals(readNameThree)) {
record.setAttribute(readTwoBarcodeTag, "readTwo2");
}
}
}
tester.addArg("BARCODE_TAG=" + barcodeTag);
tester.addArg("READ_ONE_BARCODE_TAG=" + readOneBarcodeTag);
tester.addArg("READ_TWO_BARCODE_TAG=" + readTwoBarcodeTag);
tester.runTest();
}