Examples with Chunk htsjdk.samtools.Chunk used on opensource projects

Example 1 with Chunk

use of htsjdk.samtools.Chunk in project gatk by broadinstitute.

the class NioBam method getChunksBalanced.

private static List<QueryInterval> getChunksBalanced(SamReader bam, int sequenceIndex, int retCount) {
    List<QueryInterval> ret = new ArrayList<>();
    BAMIndex index = bam.indexing().getIndex();
    SAMFileHeader header = bam.getFileHeader();
    SAMSequenceRecord s = header.getSequence(sequenceIndex);
    long totalLength = chunksLength(getChunks(index, sequenceIndex, 1, s.getSequenceLength() + 1));
    if (totalLength == 0) {
        return ret;
    }
    int sofar = 0;
    long targetLength = totalLength / retCount;
    int end = s.getSequenceLength();
    int step = s.getSequenceLength() / (100 * retCount);
    if (step < 1)
        step = 1;
    int start = 1;
    for (int j = step; j < end; j += step) {
        if (j > end)
            j = end;
        List<Chunk> candidate = getChunks(index, sequenceIndex, start, j);
        long size = chunksLength(candidate);
        if (size < targetLength) {
            // not big enough yet
            continue;
        }
        if (size > targetLength * 2) {
        // too large, search for a good separation point
        // TODO
        }
        // good, emit.
        ret.add(new QueryInterval(sequenceIndex, start, j + 1));
        start = j;
        sofar += size;
        if (ret.size() < retCount) {
            targetLength = (totalLength - sofar) / (retCount - ret.size());
        } else {
            targetLength = totalLength / retCount;
        }
    }
    return ret;
}

Example 2 with Chunk

use of htsjdk.samtools.Chunk in project hmftools by hartwigmedical.

the class BamSlicerApplication method sliceFromURLs.

private static void sliceFromURLs(@NotNull final URL indexUrl, @NotNull final URL bamUrl, @NotNull final CommandLine cmd) throws IOException {
    final File indexFile = downloadIndex(indexUrl);
    indexFile.deleteOnExit();
    final SamReader reader = SamReaderFactory.makeDefault().open(SamInputResource.of(bamUrl).index(indexFile));
    final SAMFileWriter writer = new SAMFileWriterFactory().setCreateIndex(true).makeBAMWriter(reader.getFileHeader(), true, new File(cmd.getOptionValue(OUTPUT)));
    final BAMIndex bamIndex = new DiskBasedBAMFileIndex(indexFile, reader.getFileHeader().getSequenceDictionary(), false);
    final Optional<Pair<QueryInterval[], BAMFileSpan>> queryIntervalsAndSpan = queryIntervalsAndSpan(reader, bamIndex, cmd);
    final Optional<Chunk> unmappedChunk = getUnmappedChunk(bamIndex, HttpUtils.getHeaderField(bamUrl, "Content-Length"), cmd);
    final List<Chunk> sliceChunks = sliceChunks(queryIntervalsAndSpan, unmappedChunk);
    final SamReader cachingReader = createCachingReader(indexFile, bamUrl, cmd, sliceChunks);
    queryIntervalsAndSpan.ifPresent(pair -> {
        LOGGER.info("Slicing bam on bed regions...");
        final CloseableIterator<SAMRecord> bedIterator = getIterator(cachingReader, pair.getKey(), pair.getValue().toCoordinateArray());
        writeToSlice(writer, bedIterator);
        LOGGER.info("Done writing bed slices.");
    });
    unmappedChunk.ifPresent(chunk -> {
        LOGGER.info("Slicing unmapped reads...");
        final CloseableIterator<SAMRecord> unmappedIterator = cachingReader.queryUnmapped();
        writeToSlice(writer, unmappedIterator);
        LOGGER.info("Done writing unmapped reads.");
    });
    reader.close();
    writer.close();
    cachingReader.close();
}

Example 3 with Chunk

use of htsjdk.samtools.Chunk in project hmftools by hartwigmedical.

the class BamSlicerApplication method expandChunks.

@NotNull
private static List<Chunk> expandChunks(@NotNull final List<Chunk> chunks) {
    final List<Chunk> result = Lists.newArrayList();
    for (final Chunk chunk : chunks) {
        final long chunkEndBlockAddress = BlockCompressedFilePointerUtil.getBlockAddress(chunk.getChunkEnd());
        final long extendedEndBlockAddress = chunkEndBlockAddress + BlockCompressedStreamConstants.MAX_COMPRESSED_BLOCK_SIZE;
        final long newChunkEnd = extendedEndBlockAddress > MAX_BLOCK_ADDRESS ? MAX_BLOCK_ADDRESS : extendedEndBlockAddress;
        final long chunkEndVirtualPointer = newChunkEnd << 16;
        result.add(new Chunk(chunk.getChunkStart(), chunkEndVirtualPointer));
    }
    return result;
}

Example 4 with Chunk

use of htsjdk.samtools.Chunk in project jvarkit by lindenb.

the class Biostar172515 method doWork.

@Override
public int doWork(final List<String> inputFiles) {
    final SamReaderFactory samReaderFactory = SamReaderFactory.makeDefault().setOption(SamReaderFactory.Option.CACHE_FILE_BASED_INDEXES, Boolean.TRUE).validationStringency(ValidationStringency.LENIENT);
    OutputStream stream = null;
    SamReader samReader = null;
    Set<String> args = IOUtils.unrollFiles(inputFiles);
    try {
        stream = super.openFileOrStdoutAsStream(this.outputFile);
        XMLOutputFactory xof = XMLOutputFactory.newFactory();
        this.w = xof.createXMLStreamWriter(stream);
        this.w.writeStartDocument("UTF-8", "1.0");
        this.w.writeStartElement("bai-list");
        for (final String filename : args) {
            this.w.writeStartElement("bam");
            this.w.writeAttribute("bam", filename);
            samReader = samReaderFactory.open(SamInputResource.of(filename));
            this.w.writeAttribute("has-index", String.valueOf(samReader.hasIndex()));
            if (!samReader.hasIndex()) {
                this.w.writeEndElement();
                samReader.close();
                continue;
            }
            final SamReader.Indexing indexing = samReader.indexing();
            if (!indexing.hasBrowseableIndex()) {
                this.w.writeComment("no browseable index");
                this.w.writeEndElement();
                samReader.close();
                continue;
            }
            final SAMSequenceDictionary dict = samReader.getFileHeader().getSequenceDictionary();
            this.w.writeAttribute("n_ref", String.valueOf(dict.size()));
            final BrowseableBAMIndex baiFile;
            try {
                baiFile = indexing.getBrowseableIndex();
            } catch (Exception err) {
                this.w.writeComment("no browseable index");
                this.w.writeEndElement();
                samReader.close();
                continue;
            }
            for (int tid = 0; tid < dict.size(); ++tid) {
                final SAMSequenceRecord ssr = dict.getSequence(tid);
                final BAMIndexMetaData baiMetaData = baiFile.getMetaData(tid);
                this.w.writeStartElement("reference");
                this.w.writeAttribute("ref-id", String.valueOf(tid));
                this.w.writeAttribute("ref-name", ssr.getSequenceName());
                this.w.writeAttribute("ref-length", String.valueOf(ssr.getSequenceLength()));
                this.w.writeAttribute("n_aligned", String.valueOf(baiMetaData.getAlignedRecordCount()));
                BinList binList = baiFile.getBinsOverlapping(tid, 1, ssr.getSequenceLength());
                int n_bin = 0;
                for (@SuppressWarnings("unused") final Bin binItem : binList) n_bin++;
                this.w.writeAttribute("n_bin", String.valueOf(n_bin));
                this.w.writeAttribute("n_no_coor", String.valueOf(baiMetaData.getUnalignedRecordCount()));
                for (final Bin binItem : binList) {
                    this.w.writeStartElement("bin");
                    this.w.writeAttribute("first-locus", String.valueOf(baiFile.getFirstLocusInBin(binItem)));
                    this.w.writeAttribute("last-locus", String.valueOf(baiFile.getLastLocusInBin(binItem)));
                    this.w.writeAttribute("level", String.valueOf(baiFile.getLevelForBin(binItem)));
                    final BAMFileSpan span = baiFile.getSpanOverlapping(binItem);
                    this.w.writeAttribute("first-offset", String.valueOf(span.getFirstOffset()));
                    final List<Chunk> chunks = span.getChunks();
                    this.w.writeAttribute("n_chunk", String.valueOf(chunks.size()));
                    for (final Chunk chunk : chunks) {
                        this.w.writeEmptyElement("chunk");
                        this.w.writeAttribute("chunk_beg", String.valueOf(chunk.getChunkStart()));
                        this.w.writeAttribute("chunk_end", String.valueOf(chunk.getChunkEnd()));
                    }
                    this.w.writeEndElement();
                }
                this.w.writeEndElement();
            }
            this.w.writeEndElement();
            samReader.close();
        }
        this.w.writeEndElement();
        this.w.flush();
        this.w.close();
        return 0;
    } catch (final Exception err) {
        LOG.error(err);
        return -1;
    } finally {
        CloserUtil.close(this.w);
        CloserUtil.close(stream);
        CloserUtil.close(samReader);
        this.w = null;
    }
}