@Test
public void test2() throws IOException {
SAMFileReader r =
new SAMFileReader(
new File("c:/temp/HG00096.mapped.illumina.mosaik.GBR.exome.20110411.chr20.bam"));
SAMRecordIterator iterator = r.iterator();
CompressionHeaderFactory.HuffmanParamsCalculator c = new HuffmanParamsCalculator();
String[] names = new String[100000];
for (int i = 0; i < names.length && iterator.hasNext(); i++) {
names[i] = iterator.next().getReadName();
c.add(names[i].length());
}
iterator.close();
r.close();
c.calculate();
int[] values = c.values();
int[] lens = c.bitLens();
System.out.println(Arrays.toString(values));
System.out.println(Arrays.toString(lens));
EncodingParams params = HuffmanIntegerEncoding.toParam(values, lens);
HuffmanIntegerEncoding e = new HuffmanIntegerEncoding();
e.fromByteArray(params.params);
BitCodec<Integer> codec = e.buildCodec(null, null);
ByteArrayOutputStream baos = new ByteArrayOutputStream();
DefaultBitOutputStream bos = new DefaultBitOutputStream(baos);
for (int i = 0; i < names.length; i++) {
codec.write(bos, names[i].length());
}
bos.close();
codec = e.buildCodec(null, null);
ByteArrayInputStream bais = new ByteArrayInputStream(baos.toByteArray());
DefaultBitInputStream bis = new DefaultBitInputStream(bais);
for (int i = 0; i < names.length; i++) {
int v = codec.read(bis);
if (v != names[i].length()) fail("Mismatch: " + v + " vs " + names[i].length());
}
}
@Override
public int doWork(String[] args) {
File refFile = null;
com.github.lindenb.jvarkit.util.cli.GetOpt getopt =
new com.github.lindenb.jvarkit.util.cli.GetOpt();
int c;
while ((c = getopt.getopt(args, "hvL:r:")) != -1) {
switch (c) {
case 'h':
printUsage();
return 0;
case 'v':
System.out.println(getVersion());
return 0;
case 'L':
getLogger().setLevel(java.util.logging.Level.parse(getopt.getOptArg()));
break;
case 'r':
refFile = new File(getopt.getOptArg());
break;
case ':':
System.err.println("Missing argument for option -" + getopt.getOptOpt());
return -1;
default:
System.err.println("Unknown option -" + getopt.getOptOpt());
return -1;
}
}
if (refFile == null) {
error("Undefined REF file");
return -1;
}
File bamFile = null;
if (getopt.getOptInd() + 1 != args.length) {
info("reading from stdin.");
} else {
bamFile = new File(args[getopt.getOptInd()]);
}
IndexedFastaSequenceFile indexedFastaSequenceFile = null;
SAMFileReader samFileReader = null;
try {
GenomicSequence genomicSequence = null;
indexedFastaSequenceFile = new IndexedFastaSequenceFile(refFile);
SAMFileReader.setDefaultValidationStringency(ValidationStringency.SILENT);
samFileReader = null;
if (bamFile == null) {
samFileReader = new SAMFileReader(System.in);
} else {
samFileReader = new SAMFileReader(bamFile);
}
XMLOutputFactory xmlfactory = XMLOutputFactory.newInstance();
XMLStreamWriter w = xmlfactory.createXMLStreamWriter(System.out, "UTF-8");
w.writeStartDocument("UTF-8", "1.0");
w.writeStartElement("sam");
w.writeComment(getProgramCommandLine());
w.writeAttribute("ref", (bamFile == null ? "stdin" : bamFile.getPath()));
w.writeAttribute("bam", args[1]);
SAMRecordIterator iter = samFileReader.iterator();
while (iter.hasNext()) {
SAMRecord rec = iter.next();
final byte readbases[] = rec.getReadBases();
w.writeStartElement("read");
w.writeStartElement("name");
w.writeCharacters(rec.getReadName());
w.writeEndElement();
w.writeStartElement("sequence");
w.writeCharacters(new String(readbases));
w.writeEndElement();
w.writeStartElement("flags");
w.writeAttribute("paired", String.valueOf(rec.getReadPairedFlag()));
w.writeAttribute(
"failsVendorQual", String.valueOf(rec.getReadFailsVendorQualityCheckFlag()));
w.writeAttribute("mapped", String.valueOf(!rec.getReadUnmappedFlag()));
w.writeAttribute("strand", (rec.getReadNegativeStrandFlag() ? "-" : "+"));
if (rec.getReadPairedFlag()) {
w.writeAttribute("mate-mapped", String.valueOf(!rec.getMateUnmappedFlag()));
w.writeAttribute("mate-strand", (rec.getMateNegativeStrandFlag() ? "-" : "+"));
w.writeAttribute("proper-pair", String.valueOf(rec.getProperPairFlag()));
}
w.writeCharacters(String.valueOf(rec.getFlags()));
w.writeEndElement();
if (!rec.getReadUnmappedFlag()) {
w.writeStartElement("qual");
w.writeCharacters(String.valueOf(rec.getMappingQuality()));
w.writeEndElement();
w.writeStartElement("chrom");
w.writeAttribute("index", String.valueOf(rec.getReferenceIndex()));
w.writeCharacters(rec.getReferenceName());
w.writeEndElement();
w.writeStartElement("pos");
w.writeCharacters(String.valueOf(rec.getAlignmentStart()));
w.writeEndElement();
w.writeStartElement("cigar");
w.writeCharacters(rec.getCigarString());
w.writeEndElement();
}
if (!rec.getMateUnmappedFlag()) {
w.writeStartElement("mate-chrom");
w.writeAttribute("index", String.valueOf(rec.getMateReferenceIndex()));
w.writeCharacters(rec.getMateReferenceName());
w.writeEndElement();
w.writeStartElement("mate-pos");
w.writeCharacters(String.valueOf(rec.getMateAlignmentStart()));
w.writeEndElement();
}
if (!rec.getReadUnmappedFlag()) {
if (genomicSequence == null
|| genomicSequence.getChrom().equals(rec.getReferenceName())) {
genomicSequence = new GenomicSequence(indexedFastaSequenceFile, rec.getReferenceName());
}
w.writeStartElement("align");
int readIndex = 0;
int refIndex = rec.getAlignmentStart();
for (final CigarElement e : rec.getCigar().getCigarElements()) {
switch (e.getOperator()) {
case H:
break; // ignore hard clips
case P:
break; // ignore pads
case I: // cont.
case S:
{
final int length = e.getLength();
for (int i = 0; i < length; ++i) {
w.writeEmptyElement(e.getOperator().name());
w.writeAttribute("read-index", String.valueOf(readIndex + 1));
if (readIndex >= 0 && readIndex < readbases.length) {
w.writeAttribute("read-base", String.valueOf((char) (readbases[readIndex])));
}
readIndex++;
}
break;
}
case N: // cont. -- reference skip
case D:
{
final int length = e.getLength();
for (int i = 0; i < length; ++i) {
w.writeEmptyElement(e.getOperator().name());
w.writeAttribute("ref-index", String.valueOf(refIndex));
if (refIndex >= 1 && refIndex <= genomicSequence.length()) {
w.writeAttribute(
"ref-base", String.valueOf(genomicSequence.charAt(refIndex - 1)));
}
refIndex++;
}
break;
}
case M:
case EQ:
case X:
{
final int length = e.getLength();
for (int i = 0; i < length; ++i) {
w.writeEmptyElement(e.getOperator().name());
char baseRead = '\0';
if (readIndex >= 0 && readIndex < readbases.length) {
baseRead = (char) (rec.getReadBases()[readIndex]);
w.writeAttribute("read-index", String.valueOf(readIndex + 1));
w.writeAttribute("read-base", String.valueOf(baseRead));
}
w.writeAttribute("ref-index", String.valueOf(refIndex));
if (refIndex >= 1 && refIndex <= genomicSequence.length()) {
char baseRef = genomicSequence.charAt(refIndex - 1);
w.writeAttribute("ref-base", String.valueOf(baseRef));
if (Character.toUpperCase(baseRef) != Character.toUpperCase(baseRead)) {
w.writeAttribute("mismatch", "true");
}
}
refIndex++;
readIndex++;
}
break;
}
default:
throw new IllegalStateException(
"Case statement didn't deal with cigar op: " + e.getOperator());
}
}
}
w.writeEndElement();
w.writeEndElement();
iter.close();
w.writeEndElement();
}
w.writeEndElement();
w.writeEndDocument();
w.flush();
w.close();
} catch (Exception err) {
error(err);
return -1;
} finally {
CloserUtil.close(samFileReader);
CloserUtil.close(indexedFastaSequenceFile);
}
return 0;
}
public void run(URL url) throws SocketException, IOException {
FTPSeekableStream fss = new FTPSeekableStream(url);
File index = fss.getIndexFile();
SAMFileReader reader = new SAMFileReader(fss, index, false);
reader.getFileHeader();
reader.setValidationStringency(ValidationStringency.SILENT);
logger.info("attributes");
for (Map.Entry<String, String> entry : reader.getFileHeader().getAttributes()) {
logger.info(String.format("%s : %s", entry.getKey(), entry.getValue()));
}
logger.info("sequences");
for (SAMSequenceRecord ssr : reader.getFileHeader().getSequenceDictionary().getSequences()) {
logger.info(String.format("%s : %s ", ssr.getSequenceName(), ssr.getSequenceLength()));
}
logger.info("sequences");
for (SAMSequenceRecord ssr : reader.getFileHeader().getSequenceDictionary().getSequences()) {
// if (! ssr.getSequenceName().equals("NT_166325")) {
// continue;
// }
int length = ssr.getSequenceLength();
int min = 1000;
int max = 100000;
if (min >= length) {
min = 0;
}
// logger.warn((max >= length));
if (max >= length) {
max = length;
}
if (min >= max) {
min = 0;
}
logger.info(
String.format("Sequence: %s (%s) %s-%s", ssr.getSequenceName(), length, min, max));
SAMRecordIterator i = reader.query(ssr.getSequenceName(), min, max, false);
while (i.hasNext()) {
SAMRecord record = i.next();
logger.info(
String.format(
"Read: %s (%s (%s-%s) %s) / %s",
record.getReadName(),
min,
record.getAlignmentStart(),
record.getAlignmentEnd(),
max,
record.getFlags()));
assertTrue(record.getAlignmentStart() >= min);
}
i.close();
logger.info("_________________________________________________");
}
logger.info("Done");
}
@Override
public int doWork(String[] args) {
boolean repair_missing_read = false;
SortingCollectionFactory<MappedFastq> sortingFactory =
new SortingCollectionFactory<MappedFastq>();
File forwardFile = null;
File reverseFile = null;
com.github.lindenb.jvarkit.util.cli.GetOpt opt =
new com.github.lindenb.jvarkit.util.cli.GetOpt();
int c;
sortingFactory.setComponentType(MappedFastq.class);
sortingFactory.setCodec(new MappedFastqCodec());
sortingFactory.setComparator(new MappedFastqComparator());
while ((c = opt.getopt(args, super.getGetOptDefault() + "F:R:N:r")) != -1) {
switch (c) {
case 'F':
forwardFile = new File(opt.getOptArg());
break;
case 'R':
reverseFile = new File(opt.getOptArg());
break;
case 't':
addTmpDirectory(new File(opt.getOptArg()));
break;
case 'N':
sortingFactory.setMaxRecordsInRAM(Math.max(Integer.parseInt(opt.getOptArg()), 100));
break;
case 'r':
repair_missing_read = true;
break;
case ':':
System.err.println("Missing argument for option -" + opt.getOptOpt());
return -1;
default:
{
switch (handleOtherOptions(c, opt, args)) {
case EXIT_FAILURE:
return -1;
case EXIT_SUCCESS:
return 0;
default:
break;
}
}
}
}
SAMFileReader sfr = null;
SortingCollection<MappedFastq> fastqCollection = null;
try {
sortingFactory.setTmpDirs(this.getTmpDirectories());
fastqCollection = sortingFactory.make();
fastqCollection.setDestructiveIteration(true);
boolean found_single = false;
boolean found_paired = false;
long non_primary_alignmaned_flag = 0L;
if (opt.getOptInd() == args.length) {
info("Reading from stdin");
sfr = new SAMFileReader(System.in);
} else if (opt.getOptInd() + 1 == args.length) {
String filename = args[opt.getOptInd()];
sfr = new SAMFileReader(new File(filename));
} else {
error(getMessageBundle("illegal.number.of.arguments"));
return -1;
}
sfr.setValidationStringency(ValidationStringency.LENIENT);
SAMRecordIterator iter = sfr.iterator();
SAMSequenceDictionaryProgress progress =
new SAMSequenceDictionaryProgress(sfr.getFileHeader().getSequenceDictionary());
while (iter.hasNext()) {
SAMRecord rec = iter.next();
progress.watch(rec);
if (rec.isSecondaryOrSupplementary()) {
if (non_primary_alignmaned_flag == 0) {
warning("SKIPPING NON-PRIMARY " + (non_primary_alignmaned_flag + 1) + " ALIGNMENTS");
}
non_primary_alignmaned_flag++;
continue;
}
MappedFastq m = new MappedFastq();
m.name = rec.getReadName();
if (m.name == null) m.name = "";
m.hash = m.name.hashCode();
m.seq = rec.getReadString();
if (m.seq.equals(SAMRecord.NULL_SEQUENCE_STRING)) m.seq = "";
m.qual = rec.getBaseQualityString();
if (m.qual.equals(SAMRecord.NULL_QUALS_STRING)) m.qual = "";
if (!rec.getReadUnmappedFlag() && rec.getReadNegativeStrandFlag()) {
m.seq = AcidNucleics.reverseComplement(m.seq);
m.qual = new StringBuilder(m.qual).reverse().toString();
}
if (m.seq.length() != m.qual.length()) {
error("length(seq)!=length(qual) in " + m.name);
continue;
}
if (m.seq.isEmpty() && m.qual.isEmpty()) {
m.seq = "N";
m.qual = "#";
}
if (rec.getReadPairedFlag()) {
found_paired = true;
if (found_single) {
sfr.close();
throw new PicardException("input is a mix of paired/singled reads");
}
m.side = (byte) (rec.getSecondOfPairFlag() ? 2 : 1);
} else {
found_single = true;
if (found_paired) {
sfr.close();
throw new PicardException("input is a mix of paired/singled reads");
}
m.side = (byte) 0;
}
fastqCollection.add(m);
}
iter.close();
CloserUtil.close(iter);
CloserUtil.close(sfr);
progress.finish();
fastqCollection.doneAdding();
info("Done reading.");
if (found_paired) {
FastqWriter fqw1 = null;
FastqWriter fqw2 = null;
if (forwardFile != null) {
info("Writing to " + forwardFile);
fqw1 = new BasicFastqWriter(forwardFile);
} else {
info("Writing to stdout");
fqw1 = new BasicFastqWriter(new PrintStream(System.out));
}
if (reverseFile != null) {
info("Writing to " + reverseFile);
fqw2 = new BasicFastqWriter(reverseFile);
} else {
info("Writing to interlaced stdout");
fqw2 = fqw1;
}
List<MappedFastq> row = new ArrayList<MappedFastq>();
CloseableIterator<MappedFastq> r = fastqCollection.iterator();
for (; ; ) {
MappedFastq curr = null;
if (r.hasNext()) curr = r.next();
if (curr == null || (!row.isEmpty() && !row.get(0).name.equals(curr.name))) {
if (!row.isEmpty()) {
if (row.size() > 2) {
warning("WTF :" + row);
}
boolean found_F = false;
boolean found_R = false;
for (MappedFastq m : row) {
switch ((int) m.side) {
case 1:
if (found_F)
throw new PicardException("two forward reads found for " + row.get(0).name);
found_F = true;
echo(fqw1, m);
break;
case 2:
if (found_R)
throw new PicardException("two reverse reads found for " + row.get(0).name);
found_R = true;
echo(fqw2, m);
break;
default:
throw new IllegalStateException("uh???");
}
}
if (!found_F) {
if (repair_missing_read) {
warning("forward not found for " + row.get(0));
MappedFastq pad = new MappedFastq();
pad.side = (byte) 1;
pad.name = row.get(0).name;
pad.seq = "N";
pad.qual = "#";
echo(fqw1, pad);
} else {
throw new PicardException("forward not found for " + row);
}
}
if (!found_R) {
if (repair_missing_read) {
warning("reverse not found for " + row.get(0));
MappedFastq pad = new MappedFastq();
pad.side = (byte) 2;
pad.name = row.get(0).name;
pad.seq = "N";
pad.qual = "#";
echo(fqw2, pad);
} else {
throw new PicardException("reverse not found for " + row);
}
}
}
if (curr == null) break;
row.clear();
}
row.add(curr);
}
r.close();
fqw1.close();
fqw2.close();
} else if (found_single) {
FastqWriter fqw1 = null;
if (forwardFile != null) {
info("Writing to " + forwardFile);
fqw1 = new BasicFastqWriter(forwardFile);
} else {
info("Writing to stdout");
fqw1 = new BasicFastqWriter(new PrintStream(System.out));
}
CloseableIterator<MappedFastq> r = fastqCollection.iterator();
while (r.hasNext()) {
echo(fqw1, r.next());
}
r.close();
fqw1.close();
}
return 0;
} catch (Exception err) {
error(err);
return -1;
} finally {
if (fastqCollection != null) fastqCollection.cleanup();
}
}