/** Combines multiple SAM/BAM files into one. */
@Override
protected int doWork() {
boolean matchedSortOrders = true;
// read interval list if it is defined
final List<Interval> intervalList =
(INTERVALS == null ? null : IntervalList.fromFile(INTERVALS).uniqued().getIntervals());
// map reader->iterator used if INTERVALS is defined
final Map<SamReader, CloseableIterator<SAMRecord>> samReaderToIterator =
new HashMap<SamReader, CloseableIterator<SAMRecord>>(INPUT.size());
// Open the files for reading and writing
final List<SamReader> readers = new ArrayList<SamReader>();
final List<SAMFileHeader> headers = new ArrayList<SAMFileHeader>();
{
SAMSequenceDictionary dict = null; // Used to try and reduce redundant SDs in memory
for (final File inFile : INPUT) {
IOUtil.assertFileIsReadable(inFile);
final SamReader in =
SamReaderFactory.makeDefault().referenceSequence(REFERENCE_SEQUENCE).open(inFile);
if (INTERVALS != null) {
if (!in.hasIndex())
throw new PicardException(
"Merging with interval but Bam file is not indexed " + inFile);
final CloseableIterator<SAMRecord> samIterator =
new SamRecordIntervalIteratorFactory()
.makeSamRecordIntervalIterator(in, intervalList, true);
samReaderToIterator.put(in, samIterator);
}
readers.add(in);
headers.add(in.getFileHeader());
// A slightly hackish attempt to keep memory consumption down when merging multiple files
// with
// large sequence dictionaries (10,000s of sequences). If the dictionaries are identical,
// then
// replace the duplicate copies with a single dictionary to reduce the memory footprint.
if (dict == null) {
dict = in.getFileHeader().getSequenceDictionary();
} else if (dict.equals(in.getFileHeader().getSequenceDictionary())) {
in.getFileHeader().setSequenceDictionary(dict);
}
matchedSortOrders = matchedSortOrders && in.getFileHeader().getSortOrder() == SORT_ORDER;
}
}
// If all the input sort orders match the output sort order then just merge them and
// write on the fly, otherwise setup to merge and sort before writing out the final file
IOUtil.assertFileIsWritable(OUTPUT);
final boolean presorted;
final SAMFileHeader.SortOrder headerMergerSortOrder;
final boolean mergingSamRecordIteratorAssumeSorted;
if (matchedSortOrders
|| SORT_ORDER == SAMFileHeader.SortOrder.unsorted
|| ASSUME_SORTED
|| INTERVALS != null) {
log.info(
"Input files are in same order as output so sorting to temp directory is not needed.");
headerMergerSortOrder = SORT_ORDER;
mergingSamRecordIteratorAssumeSorted = ASSUME_SORTED;
presorted = true;
} else {
log.info("Sorting input files using temp directory " + TMP_DIR);
headerMergerSortOrder = SAMFileHeader.SortOrder.unsorted;
mergingSamRecordIteratorAssumeSorted = false;
presorted = false;
}
final SamFileHeaderMerger headerMerger =
new SamFileHeaderMerger(headerMergerSortOrder, headers, MERGE_SEQUENCE_DICTIONARIES);
final MergingSamRecordIterator iterator;
// no interval defined, get an iterator for the whole bam
if (intervalList == null) {
iterator =
new MergingSamRecordIterator(headerMerger, readers, mergingSamRecordIteratorAssumeSorted);
} else {
// show warning related to https://github.com/broadinstitute/picard/pull/314/files
log.info(
"Warning: merged bams from different interval lists may contain the same read in both files");
iterator = new MergingSamRecordIterator(headerMerger, samReaderToIterator, true);
}
final SAMFileHeader header = headerMerger.getMergedHeader();
for (final String comment : COMMENT) {
header.addComment(comment);
}
header.setSortOrder(SORT_ORDER);
final SAMFileWriterFactory samFileWriterFactory = new SAMFileWriterFactory();
if (USE_THREADING) {
samFileWriterFactory.setUseAsyncIo(true);
}
final SAMFileWriter out = samFileWriterFactory.makeSAMOrBAMWriter(header, presorted, OUTPUT);
// Lastly loop through and write out the records
final ProgressLogger progress = new ProgressLogger(log, PROGRESS_INTERVAL);
while (iterator.hasNext()) {
final SAMRecord record = iterator.next();
out.addAlignment(record);
progress.record(record);
}
log.info("Finished reading inputs.");
for (final CloseableIterator<SAMRecord> iter : samReaderToIterator.values())
CloserUtil.close(iter);
CloserUtil.close(readers);
out.close();
return 0;
}
