Java AsciiFeatureCodec类代码示例

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
private void makeTabixCompressedIndex(final File sourceFile, final File indexFile, final AsciiFeatureCodec codec,
                                      final TabixFormat format) throws IOException {
    TabixIndexCreator indexCreator = new TabixIndexCreator(format);

    try (
        BlockCompressedInputStream inputStream = new BlockCompressedInputStream(
            new FileInputStream(sourceFile));
        LittleEndianOutputStream outputStream = new LittleEndianOutputStream(
            new BlockCompressedOutputStream(indexFile))
    ) {
        long p = 0;
        String line = inputStream.readLine();

        while (line != null) {
            //add the feature to the index
            Feature decode = codec.decode(line);
            if (decode != null) {
                indexCreator.addFeature(decode, p);
            }
            // read the next line if available
            p = inputStream.getFilePointer();
            line = inputStream.readLine();
        }

        // write the index to a file
        Index index = indexCreator.finalizeIndex(p);
        // VERY important! either use write based on input file or pass the little endian a BGZF stream
        index.write(outputStream);
    }
}
 

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
private void createGeneCompressedIndex(File indexFile, File file, GffCodec.GffType gffType) throws IOException {
    AsciiFeatureCodec<GeneFeature> codec = new GffCodec(gffType);
    TabixIndexCreator indexCreator = new TabixIndexCreator(TabixFormat.GFF);

    try (
        BlockCompressedInputStream inputStream = new BlockCompressedInputStream(new FileInputStream(file));
        LittleEndianOutputStream outputStream = new LittleEndianOutputStream(
            new BlockCompressedOutputStream(indexFile))
    ) {
        long p = 0;
        String line = inputStream.readLine();

        while (line != null) {
            //add the feature to the index
            GeneFeature decode = codec.decode(line);
            if (decode != null) {
                indexCreator.addFeature(decode, p);
            }
            // read the next line if available
            p = inputStream.getFilePointer();
            line = inputStream.readLine();
        }

        // write the index to a file
        Index index = indexCreator.finalizeIndex(p);
        // VERY important! either use write based on input file or pass the little endian a BGZF stream
        index.write(outputStream);
    }
}
 

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
private GeneFile registerGeneFileFromUrl(final FeatureIndexedFileRegistrationRequest request) {
    Reference reference = referenceGenomeManager.loadReferenceGenome(request.getReferenceId());
    Map<String, Chromosome> chromosomeMap = Utils.makeChromosomeMap(reference);

    GeneFile geneFile = null;

    final String extension = Utils.getFileExtension(request.getPath());
    GffCodec.GffType gffType = GffCodec.GffType.forExt(extension);
    AsciiFeatureCodec<GeneFeature> codec = new GffCodec(gffType);

    try (FeatureReader<GeneFeature> reader = AbstractFeatureReader.getFeatureReader(request.getPath(),
            request.getIndexPath(), codec, true)) {
        geneFile = createGeneFile(request);
        boolean hasGenes = false;
        for (Map.Entry<String, Chromosome> chrEntry : chromosomeMap.entrySet()) {
            CloseableIterator<GeneFeature> iterator = Utils.query(reader, chrEntry.getKey(), 1,
                    chrEntry.getValue().getSize());
            if (iterator.hasNext()) {
                hasGenes = true;
                break;
            }
        }

        Assert.isTrue(hasGenes, getMessage(MessagesConstants.ERROR_FILE_CORRUPTED_OR_EMPTY, request.getPath(),
                reference.getName()));

        biologicalDataItemManager.createBiologicalDataItem(geneFile.getIndex());
        geneFileManager.createGeneFile(geneFile);
    }  catch (IOException e) {
        throw new RegistrationException(getMessage(ERROR_REGISTER_FILE, request.getName()), e);
    } finally {
        if (geneFile != null && geneFile.getId() != null &&
                !geneFileManager.geneFileExists(geneFile.getId())) {
            biologicalDataItemManager.deleteBiologicalDataItem(geneFile.getBioDataItemId());
        }
    }

    return geneFile;
}
 

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
private void openStreams(GeneFile geneFile, IndexedFileRegistrationRequest request)
    throws
    IOException {
    File file = new File(request.getPath());

    final String extension = Utils.getFileExtension(request.getPath());
    GffCodec.GffType gffType = GffCodec.GffType.forExt(extension);
    AsciiFeatureCodec<GeneFeature> codec = new GffCodec(gffType);

    compressedInputStream = new BlockCompressedInputStream(new FileInputStream(file));
    inputStream = new PositionalBufferedStream(new FileInputStream(file));

    if (geneFile.getCompressed()) {
        largeScaleOS = null;
        largeScaleBCOS = fileManager.makeGeneBlockCompressedOutputStream(gffType, geneFile,
                                                                         GeneFileType.LARGE_SCALE);
        transcriptOS = null;
        transcriptBCOS = fileManager.makeGeneBlockCompressedOutputStream(gffType, geneFile,
                                                                         GeneFileType.TRANSCRIPT);
        writerTranscript = new BufferedWriter(new OutputStreamWriter(transcriptBCOS,
                                                                     Charset.forName(CHARSET_NAME)));
        writerLargeScale = new BufferedWriter(new OutputStreamWriter(largeScaleBCOS,
                                                                     Charset.forName(CHARSET_NAME)));
        iterator = new IndexUtils.FeatureIterator<>(file, codec);
    } else {
        largeScaleOS = fileManager.makePositionalOutputStream(gffType, geneFile, GeneFileType.LARGE_SCALE);
        largeScaleBCOS = null;
        transcriptOS = fileManager.makePositionalOutputStream(gffType, geneFile, GeneFileType.TRANSCRIPT);
        transcriptBCOS = null;

        writerTranscript = new BufferedWriter(new OutputStreamWriter(transcriptOS, Charset.forName(CHARSET_NAME)));
        writerLargeScale = new BufferedWriter(new OutputStreamWriter(largeScaleOS, Charset.forName(CHARSET_NAME)));
        iterator = new IndexUtils.FeatureIterator<>(inputStream, codec);
    }
}
 

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
SQLCodecSource(DBProfile.DBTable table, AsciiFeatureCodec codec) {
    super(table);
    this.codec = codec;
    this.binColName = table.getBinColName();
    this.chromoColName = table.getChromoColName();
    this.posStartColName = table.getPosStartColName();
    this.posEndColName = table.getPosEndColName();
    this.startColIndex = table.getStartColIndex();
    this.endColIndex = table.getEndColIndex();

    readHeader();
}
 

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
/**
 * @param table
 * @return a SQLCodecSource, or null if no appropriate codec found
 */
public static SQLCodecSource getFromTable(DBProfile.DBTable table) {
    FeatureCodec codec = CodecFactory.getCodec("." + table.getFormat(), GenomeManager.getInstance().getCurrentGenome());
    if (codec != null && codec instanceof AsciiFeatureCodec) {
        return new SQLCodecSource(table, (AsciiFeatureCodec) codec);
    }
    return null;
}
 

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
/**
 * Return an parser instance appropriate the the file type.  Currently the filename
 * is used to determine file type, this is fragile obviously but what is the alternative?
 */
public static FeatureParser getInstanceFor(ResourceLocator locator, Genome genome) {
    FeatureCodec codec = CodecFactory.getCodec(locator, genome);
    if (codec != null && codec instanceof AsciiFeatureCodec) {
        return new FeatureCodecParser((AsciiFeatureCodec) codec, genome);
    } else {
        return null;
    }
}
 

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
public static <T extends Range> List<T> parse(AsciiFeatureCodec<T> codec, String path) throws IOException {
    List<T> values = new ArrayList<T>();
    AbstractFeatureReader reader = AbstractFeatureReader.getFeatureReader(path, codec, false);
    Iterator<T> iter = reader.iterator();
    while(iter.hasNext()){
        values.add(iter.next());
    }
    return values;
}
 

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
private void makeTabixIndex(final File sourceFile, final File indexFile,
                            final AsciiFeatureCodec codec, final TabixFormat format) throws IOException {
    TabixIndex index = IndexFactory.createTabixIndex(sourceFile, codec, format, null);
    index.write(indexFile);
}
 

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
/**
 * Creates an index for gene file, determined by a {@code GeneFile} object
 *
 * @param geneFeatureClass {@code Class<? extends GeneFeature>} defines gene file type, that will be indexed
 * @param geneFile         {@code GeneFile} that represents a file in the system
 */
public void makeGeneIndex(Class<? extends GeneFeature> geneFeatureClass, final GeneFile geneFile, final GeneFileType
        type) throws IOException {
    GffCodec.GffType gffType = GffCodec.GffType.forGeneFile(geneFeatureClass, geneFile);

    final Map<String, Object> params = new HashMap<>();
    params.put(DIR_ID.name(), geneFile.getId());
    params.put(USER_ID.name(), geneFile.getCreatedBy());
    params.put(GENE_EXTENSION.name(), gffType.getExtensions()[0]);

    File file;
    File indexFile;
    switch (type) {
        case ORIGINAL:
            file = new File(geneFile.getPath());
            indexFile = new File(toRealPath(substitute(GENE_INDEX, params)));
            break;
        case LARGE_SCALE:
            file = new File(toRealPath(substitute(GENE_LARGE_SCALE_FILE, params)));
            indexFile = new File(toRealPath(substitute(GENE_LARGE_SCALE_INDEX, params)));
            break;
        case TRANSCRIPT:
            file = new File(toRealPath(substitute(GENE_TRANSCRIPT_FILE, params)));
            indexFile = new File(toRealPath(substitute(GENE_TRANSCRIPT_INDEX, params)));
            break;
        default:
            throw new UnsupportedGeneFileTypeException(type);
    }

    LOGGER.info(getMessage(MessagesConstants.INFO_GENE_INDEX_WRITING, indexFile.getAbsolutePath()));

    if (geneFile.getCompressed()) {
        createGeneCompressedIndex(indexFile, file, gffType);
    } else {
        AsciiFeatureCodec<GeneFeature> codec = new GffCodec(gffType);
        TabixIndex index = IndexUtils.createTabixIndex(file, codec, TabixFormat.GFF);
        index.write(indexFile);
    }

    if (type.equals(GeneFileType.ORIGINAL)) {
        BiologicalDataItem indexItem = new BiologicalDataItem();
        indexItem.setCreatedDate(new Date());
        indexItem.setPath(indexFile.getAbsolutePath());
        indexItem.setFormat(BiologicalDataItemFormat.GENE_INDEX);
        indexItem.setType(BiologicalDataItemResourceType.FILE);
        indexItem.setName("");
        indexItem.setCreatedBy(AuthUtils.getCurrentUserId());

        geneFile.setIndex(indexItem);
    }
}
 

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
public VCFWrapperCodec(AsciiFeatureCodec wrappedCodec, Genome genome) {
    super(VCFVariant.class);
    this.wrappedCodec = wrappedCodec;
    this.genome = genome;
}
 

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
public FeatureCodecParser(AsciiFeatureCodec codec, Genome genome) {
    this.codec = codec;
}
 

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
public DecoderWrapper(AsciiFeatureCodec<T> wrappedCodec) {
    this.wrappedCodec = wrappedCodec;
    this.lineFeatureDecoder = this;
}
 

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
public RenameDecoder(AsciiFeatureCodec<T> featureCodec){
    super(new AsciiDecoder.DecoderWrapper<T>(featureCodec));
}
 

import htsjdk.tribble.AsciiFeatureCodec; //导入依赖的package包/类
/**
 * Create a {@code AbstractFeatureReader<GeneFeature, LineIterator>} reader for given file, optionally
 * uses an index
 *
 * @param path      a path to gene file with .gff, .gtf, .gff.gz or .gtf.gz extensions
 * @param index a path to index file whith .tbi extension, set null if no index should be used
 * @param useIndex  {@code boolean} determines if index should be used
 * @return {@code AbstractFeatureReader<GeneFeature, LineIterator>} a reader to work with gene file
 */
public AbstractFeatureReader<GeneFeature, LineIterator> makeGeneReader(String path, String index,
                                                                       boolean useIndex) {
    String extension = getGeneFileExtension(path);
    Assert.notNull(extension, getMessage(MessagesConstants.ERROR_UNSUPPORTED_GENE_FILE_EXTESION));

    AsciiFeatureCodec<GeneFeature> codec = new GffCodec(GffCodec.GffType.forExt(extension));
    return AbstractFeatureReader.getFeatureReader(path, index, codec, useIndex);
}