/**
* Asserts that files are readable and writable and then fires off an HsMetricsCalculator instance
* to do the real work.
*/
protected int doWork() {
IoUtil.assertFileIsReadable(getProbeIntervals());
IoUtil.assertFileIsReadable(TARGET_INTERVALS);
IoUtil.assertFileIsReadable(INPUT);
IoUtil.assertFileIsWritable(OUTPUT);
if (PER_TARGET_COVERAGE != null) IoUtil.assertFileIsWritable(PER_TARGET_COVERAGE);
final SAMFileReader samReader = new SAMFileReader(INPUT);
final File probeIntervals = getProbeIntervals();
// Validate that the targets and baits have the same references as the reads file
SequenceUtil.assertSequenceDictionariesEqual(
samReader.getFileHeader().getSequenceDictionary(),
IntervalList.fromFile(TARGET_INTERVALS).getHeader().getSequenceDictionary(),
INPUT,
TARGET_INTERVALS);
SequenceUtil.assertSequenceDictionariesEqual(
samReader.getFileHeader().getSequenceDictionary(),
IntervalList.fromFile(probeIntervals).getHeader().getSequenceDictionary(),
INPUT,
probeIntervals);
ReferenceSequenceFile ref = null;
if (REFERENCE_SEQUENCE != null) {
IoUtil.assertFileIsReadable(REFERENCE_SEQUENCE);
ref = ReferenceSequenceFileFactory.getReferenceSequenceFile(REFERENCE_SEQUENCE);
SequenceUtil.assertSequenceDictionariesEqual(
samReader.getFileHeader().getSequenceDictionary(),
ref.getSequenceDictionary(),
INPUT,
REFERENCE_SEQUENCE);
}
final TargetMetricsCollector collector =
makeCollector(
METRIC_ACCUMULATION_LEVEL,
samReader.getFileHeader().getReadGroups(),
ref,
PER_TARGET_COVERAGE,
TARGET_INTERVALS,
probeIntervals,
getProbeSetName());
// Add each record to the requested collectors
final Iterator<SAMRecord> records = samReader.iterator();
final ProgressLogger progress = new ProgressLogger(log);
while (records.hasNext()) {
final SAMRecord sam = records.next();
collector.acceptRecord(sam, null);
progress.record(sam);
}
// Write the output file
final MetricsFile<HsMetrics, Integer> metrics = getMetricsFile();
collector.finish();
collector.addAllLevelsToFile(metrics);
metrics.write(OUTPUT);
return 0;
}
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();
}
}