@Override
public boolean reallyFilterOut(final SAMRecord record) {
return record.getDuplicateReadFlag();
}
//
// Returns a downsampled set of reads for each sample.
//
private List<List<SAMRecord>> getReads(
List<String> inputFiles, List<Feature> regions, ReAligner realigner) {
int downsampleTarget = desiredNumberOfReads(regions);
List<DownsampledReadList> readsList = new ArrayList<DownsampledReadList>();
for (String input : inputFiles) {
Set<String> readIds = new HashSet<String>();
DownsampledReadList reads = new DownsampledReadList(downsampleTarget);
readsList.add(reads);
for (Feature region : regions) {
SAMFileReader reader = new SAMFileReader(new File(input));
reader.setValidationStringency(ValidationStringency.SILENT);
Iterator<SAMRecord> iter;
if (region != null) {
iter =
reader.queryOverlapping(
region.getSeqname(), (int) region.getStart(), (int) region.getEnd());
} else {
iter = reader.iterator();
}
while (iter.hasNext()) {
SAMRecord read = iter.next();
// Don't allow same read to be counted twice.
if ((!realigner.isFiltered(read))
&& (!read.getDuplicateReadFlag())
&& (!read.getReadFailsVendorQualityCheckFlag())
&& (read.getMappingQuality() >= realigner.getMinMappingQuality()
|| read.getReadUnmappedFlag())
&& (!readIds.contains(getIdentifier(read)))) {
if (read.getReadString().length() > readLength) {
reader.close();
throw new IllegalArgumentException(
"Maximum read length of: "
+ readLength
+ " exceeded for: "
+ read.getSAMString());
}
readIds.add(getIdentifier(read));
reads.add(read);
}
}
if (reads.getTotalReadCount() != reads.getReads().size()) {
if (isDebug) {
System.err.println(
"downsampled: "
+ regions.get(0).getDescriptor()
+ ": "
+ reads.getTotalReadCount()
+ " -> "
+ reads.getReads().size());
}
}
reader.close();
}
}
List<List<SAMRecord>> sampleReads = new ArrayList<List<SAMRecord>>();
for (DownsampledReadList downsampledReads : readsList) {
sampleReads.add(downsampledReads.getReads());
}
return sampleReads;
}
/**
* Returns true if we don't think this read is eligible for the BAQ calculation. Examples include
* non-PF reads, duplicates, or unmapped reads. Used by baqRead to determine if a read should fall
* through the calculation.
*
* @param read
* @return
*/
public boolean excludeReadFromBAQ(SAMRecord read) {
// keeping mapped reads, regardless of pairing status, or primary alignment status.
return read.getReadUnmappedFlag()
|| read.getReadFailsVendorQualityCheckFlag()
|| read.getDuplicateReadFlag();
}