Java TabixFormat.GFF属性代码示例

public static TabixFormat trackFormatToTabixFormat(TrackFormat fmt){
	
	TabixFormat tbx= null;
	if(fmt.equals(TrackFormat.BAM)){
		tbx= TabixFormat.SAM; 
	} else if (fmt.equals(TrackFormat.BED) || fmt.equals(TrackFormat.BEDGRAPH)){
		tbx= TabixFormat.BED; 
	} else if (fmt.equals(TrackFormat.GFF) || fmt.equals(TrackFormat.GTF)){
		tbx= TabixFormat.GFF;
	} else if (fmt.equals(TrackFormat.VCF)){
		tbx= TabixFormat.VCF;
	} else {
		throw new RuntimeException();
	}
	return tbx;
	
}
 

@Test
public void canCompressAndIndexSortedGzipFile() throws IOException, InvalidRecordException, ClassNotFoundException, SQLException {
	
	String infile= "test_data/hg19_genes.gtf.gz";
	File outfile= new File("test_data/tmp2.bed.gz");
	outfile.deleteOnExit();
	
	File expectedTbi= new File(outfile.getAbsolutePath() + TabixUtils.STANDARD_INDEX_EXTENSION); 
	expectedTbi.deleteOnExit();
	
	new MakeTabixIndex(infile, outfile, TabixFormat.GFF);

	assertTrue(outfile.exists());
	assertTrue(outfile.length() > 7000000);
	assertTrue(expectedTbi.exists());
	assertTrue(expectedTbi.length() > 500000);
	
	TabixReader tbx = new TabixReader(outfile.getAbsolutePath());
	Iterator x = tbx.query("chr1", 1, 1000000);
	assertTrue(x.next().startsWith("chr1"));
	
}
 

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);
    }
}
 

public TrackBookmark(GenomicCoords gc, String nameForBookmark) throws IOException, ClassNotFoundException, InvalidRecordException, SQLException, InvalidGenomicCoordsException{
	super(gc);
			
	this.setTrackTag(trackName);
			
	// Prepare bookmark file
	// =====================
    // First write out the current position as plain text. Then gzip and index.
	File bookmarkPlain= Utils.createTempFile(".asciigenome.bookmarks.", ".gff");
	bookmarkPlain.deleteOnExit();
	BufferedWriter wr = new BufferedWriter(new FileWriter(bookmarkPlain));
	wr.write(this.positionToGffLine(gc, nameForBookmark) + "\n"); // gc.getChrom() + "\t" + (gc.getFrom() - 1) + "\t" + gc.getTo() + "\t" + nameForBookmark + "\n");
	wr.close();

	File bookmark= new File(bookmarkPlain + ".gz");
	bookmark.deleteOnExit();
	(new File(bookmark.getAbsolutePath() + ".tbi")).deleteOnExit();
	this.setFilename(bookmark.getAbsolutePath());
	this.setWorkFilename(bookmark.getAbsolutePath());
	
	new MakeTabixIndex(bookmarkPlain.getAbsolutePath(), bookmark, TabixFormat.GFF);
	bookmarkPlain.delete();
	
	this.setTabixReader(new TabixReader(bookmark.getAbsolutePath()));
	this.setTrackFormat(TrackFormat.GTF);
	this.setGc(gc);
}
 

/** Add current genomic position to track.  
 * @throws IOException 
 * @throws SQLException 
 * @throws InvalidRecordException 
 * @throws ClassNotFoundException 
 * @throws InvalidGenomicCoordsException 
 * */
@Override
public void addBookmark(GenomicCoords gc, String nameForBookmark) throws IOException, ClassNotFoundException, InvalidRecordException, SQLException, InvalidGenomicCoordsException{
	
	// Adding a bookmark means re-preparing the bgzip file again. 
	
	// First write the current position to a new tmp file, then append to this file
	// all the bookmarks previously added. Then bgzip and compress replacing the old bookmark file.
	File plainNew= new File(this.getWorkFilename() + ".add");
	plainNew.deleteOnExit();
	BufferedWriter wr = new BufferedWriter(new FileWriter(plainNew));
	
	InputStream fileStream = new FileInputStream(this.getWorkFilename());
	Reader decoder = new InputStreamReader(new GZIPInputStream(fileStream), "UTF-8");
	BufferedReader br= new BufferedReader(decoder);

	String line;
	while( (line = br.readLine()) != null) {
		if(line.contains("\t__ignore_me__")){ // Hack to circumvent issue #38
			continue;
		}			
		wr.write(line + "\n");
	}
	// Add new bookamrk
	wr.write(this.positionToGffLine(gc, nameForBookmark) + "\n");
	br.close();
	wr.close();

	// Recompress and index replacing the original bgzip file
	new MakeTabixIndex(plainNew.getAbsolutePath(), new File(this.getWorkFilename()), TabixFormat.GFF);
	plainNew.delete();
	this.tabixReader= new TabixReader(this.getWorkFilename());
	// Update track.
	this.update();
}
 

/** Remove the bookmark matching the exact coordinates of the current position. 
 * Bookmarks partially overlapping are not removed. 
 * @param bookmarkRegion 
 * @throws IOException 
 * @throws SQLException 
 * @throws InvalidRecordException 
 * @throws ClassNotFoundException 
 * @throws InvalidGenomicCoordsException 
 * */
public void removeBookmark(GenomicCoords bookmarkRegion) throws IOException, ClassNotFoundException, InvalidRecordException, SQLException, InvalidGenomicCoordsException {
	// To remove a bookmark, iterate through the bgzip file writing records to a tmp file.
	// The record(s) matching this position is not written. 
	// 
	File plainNew= new File(this.getWorkFilename() + ".remove");
	plainNew.deleteOnExit();
	BufferedWriter wr = new BufferedWriter(new FileWriter(plainNew));
	
	InputStream fileStream = new FileInputStream(this.getWorkFilename());
	Reader decoder = new InputStreamReader(new GZIPInputStream(fileStream), "UTF-8");
	BufferedReader br= new BufferedReader(decoder);

	String line;
	while( (line = br.readLine()) != null) {
		
		if( ! line.startsWith("#")){ // In case of comment lines
			List<String>pos= Lists.newArrayList(Splitter.on("\t").split(line));
			
			if(bookmarkRegion.getChrom().equals(pos.get(0)) && 
			   bookmarkRegion.getFrom() == Integer.parseInt(pos.get(3)) && 
			   bookmarkRegion.getTo() == Integer.parseInt(pos.get(4))){
				continue;
			}
		}
		wr.write(line + "\n");
	}
	wr.close();
	br.close();
	
	// Recompress and index replacing the original bgzip file
	new MakeTabixIndex(plainNew.getAbsolutePath(), new File(this.getWorkFilename()), TabixFormat.GFF);
	plainNew.delete();
	this.tabixReader= new TabixReader(this.getWorkFilename());
	// Update track.
	this.update();
}
 

/**
 * Block compress input file and create associated tabix index. Newly created file and index are
 * deleted on exit if deleteOnExit true.
 * @throws IOException 
 * @throws InvalidRecordException 
 * */
private void blockCompressAndIndex(String in, String bgzfOut, boolean deleteOnExit) throws IOException, InvalidRecordException {
			
	File inFile= new File(in);
	File outFile= new File(bgzfOut);
	
	LineIterator lin= utils.IOUtils.openURIForLineIterator(inFile.getAbsolutePath());

	BlockCompressedOutputStream writer = new BlockCompressedOutputStream(outFile);
	long filePosition= writer.getFilePointer();
			
	TabixIndexCreator indexCreator=new TabixIndexCreator(TabixFormat.GFF);		
	
	while(lin.hasNext()){
		String line = lin.next();			
		GtfLine gtf= new GtfLine(line.split("\t"));
		writer.write(line.getBytes());
		writer.write('\n');
		indexCreator.addFeature(gtf, filePosition);
		filePosition = writer.getFilePointer();
	}
	writer.flush();
			
	File tbi= new File(bgzfOut + TabixUtils.STANDARD_INDEX_EXTENSION);
	if(tbi.exists() && tbi.isFile()){
		writer.close();
		throw new RuntimeException("Index file exists: " + tbi);
	}
	Index index = indexCreator.finalizeIndex(writer.getFilePointer());
	index.writeBasedOnFeatureFile(outFile);
	writer.close();
	
	if(deleteOnExit){
		outFile.deleteOnExit();
		File idx= new File(outFile.getAbsolutePath() + TabixUtils.STANDARD_INDEX_EXTENSION);
		idx.deleteOnExit();
	}
}
 

/** Fetch genePred data and convert it to tabix indexed GTF. 
 * @param maxRecords stop reading after this many records. If maxRecrods < 0 read entire data. 
 * Use this option to test whether connection is suitable genePred input.   
 * */
public UcscGenePred(String databaseTableOrFile, int maxRecords) throws IOException, InvalidCommandLineException, InvalidGenomicCoordsException, ClassNotFoundException, InvalidRecordException, SQLException{

	if(maxRecords < 0){
		maxRecords= Integer.MAX_VALUE;
	}
	
	// Prepare reader 
	File genePred= new File(databaseTableOrFile); 

	BufferedReader br; 

	if(genePred.isFile()){
		try{
			InputStream fileStream = new FileInputStream(genePred);
			Reader decoder = new InputStreamReader(new GZIPInputStream(fileStream), "UTF-8");
			br= new BufferedReader(decoder);
		} catch (Exception e){
			br= new BufferedReader(new FileReader(genePred));
		}

	} else if(Utils.urlFileExists(databaseTableOrFile)) {

			br= this.readFromUrl(databaseTableOrFile);
		
	} else {
		// Read from UCSC if not local file or remote file.
		List<String> conn= Lists.newArrayList((Splitter.on(":").omitEmptyStrings().split(databaseTableOrFile)));
		if(conn.size() != 2){
			System.err.println(databaseTableOrFile + " is not a local file and does not look like a string in the form database:table");
			throw new InvalidCommandLineException();
		}
		br= this.readGenePredFromUcsc(conn.get(0), conn.get(1));
	} 
	
	// Prepare output gtf file
	String source= (new File(databaseTableOrFile).getName()).replaceAll(".txt.gz$|.txt$", "");
	File tmpGtf= File.createTempFile("asciigenome.", ".ucsc." + source + ".gtf");
	tmpGtf.deleteOnExit();

	BufferedWriter wr= new BufferedWriter(new FileWriter(tmpGtf));
	
	String line= null;
	int i= 0;
	while((line = br.readLine()) != null && i < maxRecords){
		if(line.startsWith("#")){ // Are comment lines allowed at all?
			continue;
		}
		List<IntervalFeature> gene= this.genePredRecordToGene(line, source);
		i++;
		for(IntervalFeature x : gene){
			wr.write(x.getRaw() + "\n");
		}
	}
	if(maxRecords == Integer.MAX_VALUE){
		// System.err.println(i + " genePred records converted to " + j + " GTF features (" + Math.rint((t1-t0) / 1000) + " s).");
	}
	br.close();
	wr.close();
	
	// Tabix conversion
	String infile= tmpGtf.getAbsolutePath();
	File outfile= new File(infile + ".gz");
	outfile.deleteOnExit();
	File expectedTbi= new File(outfile.getAbsolutePath() + ".tbi"); 
	expectedTbi.deleteOnExit();
	
	new MakeTabixIndex(infile, outfile, TabixFormat.GFF);
	tmpGtf.delete();
	this.tabixFile= outfile.getAbsolutePath();
}