Examples with TrackFormat tracks.TrackFormat used on opensource projects

Example 1 with TrackFormat

use of tracks.TrackFormat in project ASCIIGenome by dariober.

the class Main method initRegion.

/**
 * Return a suitable region to start. If a region is already given, do nothing.
 * This method is a mess and should be cleaned up together with GenomicCoords class.
 * @throws InvalidGenomicCoordsException
 */
public static String initRegion(List<String> inputFileList, String fasta, String genome, int debug) throws IOException, InvalidGenomicCoordsException {
    // Preferably we start from a position that has a feature rather than from the start of a
    // random chrom.
    System.err.print("Initializing coordinates... ");
    // First search for files that can init chrom and position
    List<String> skipped = new ArrayList<String>();
    for (String x : inputFileList) {
        TrackFormat fmt = Utils.getFileTypeFromName(x);
        if (fmt.equals(TrackFormat.TDF)) {
            skipped.add(x);
            continue;
        }
        try {
            String region = Utils.initRegionFromFile(x);
            System.err.println("Done from: " + x);
            return region;
        } catch (Exception e) {
            System.err.println("\nCould not initilize from file " + x);
            if (debug > 0) {
                e.printStackTrace();
            }
        }
    }
    // Try to initialize from fasta
    if (fasta != null && !fasta.trim().isEmpty()) {
        IndexedFastaSequenceFile faSeqFile = new IndexedFastaSequenceFile(new File(fasta));
        String region = faSeqFile.nextSequence().getName();
        faSeqFile.close();
        return region;
    }
    // Try genome file
    if (genome != null && !genome.trim().isEmpty()) {
        GenomicCoords gc = new GenomicCoords(Utils.getTerminalWidth());
        gc.setGenome(Arrays.asList(new String[] { genome }), false);
        SAMSequenceDictionary samSeqDict = gc.getSamSeqDict();
        String region = samSeqDict.getSequence(0).getSequenceName();
        return region;
    }
    // Failing that, look for any file that gives at least chrom
    for (String x : skipped) {
        try {
            String region = Utils.initRegionFromFile(x);
            System.err.println("Done from: " + x);
            return region;
        } catch (Exception e) {
            System.err.println("\nCould not initilize from file " + x);
            if (debug > 0) {
                e.printStackTrace();
            }
        }
    }
    // It appears everything failed to initialise...
    return "";
}

Example 2 with TrackFormat

use of tracks.TrackFormat in project ASCIIGenome by dariober.

the class Utils method prepareStdinFile.

/**
 * Read file from stdin and write to a tmp file with name extension
 * consistent with the input format.
 * This method should be private. Set to protected only for testing.
 */
protected static File prepareStdinFile() throws IOException {
    Scanner sc = new Scanner(System.in);
    File tmp = createTempFile("stdin.", "");
    tmp.deleteOnExit();
    BufferedWriter bw = new BufferedWriter(new FileWriter(tmp));
    while (sc.hasNextLine()) {
        bw.write(sc.nextLine() + "\n");
    }
    bw.close();
    TrackFormat fmt = sniffFile(tmp);
    String fmtName;
    if (fmt.equals(TrackFormat.BAM)) {
        fmtName = tmp.getAbsoluteFile() + ".sam";
    } else if (fmt.equals(TrackFormat.VCF)) {
        fmtName = tmp.getAbsoluteFile() + ".vcf";
    } else if (fmt.equals(TrackFormat.BEDGRAPH)) {
        fmtName = tmp.getAbsoluteFile() + ".bedGraph";
    } else if (fmt.equals(TrackFormat.BED)) {
        fmtName = tmp.getAbsoluteFile() + ".bed";
    } else if (fmt.equals(TrackFormat.GFF)) {
        fmtName = tmp.getAbsoluteFile() + ".gff3";
    } else if (fmt.equals(TrackFormat.GTF)) {
        fmtName = tmp.getAbsoluteFile() + ".gtf";
    } else {
        throw new IOException("Cannot determine track format of stdin.");
    }
    File fmtFile = new File(fmtName);
    fmtFile.deleteOnExit();
    tmp.renameTo(fmtFile);
    tmp.delete();
    return fmtFile;
}

Example 3 with TrackFormat

use of tracks.TrackFormat in project ASCIIGenome by dariober.

the class Utils method initRegionFromFile.

/**
 * Get the first chrom string from first line of input file. As you add support for more filetypes you should update
 * this function. This method is very dirty and shouldn't be trusted 100%
 * @throws InvalidGenomicCoordsException
 * @throws SQLException
 * @throws InvalidRecordException
 * @throws InvalidCommandLineException
 * @throws ClassNotFoundException
 */
@SuppressWarnings("unused")
public static String initRegionFromFile(String x) throws IOException, InvalidGenomicCoordsException, ClassNotFoundException, InvalidCommandLineException, InvalidRecordException, SQLException {
    UrlValidator urlValidator = new UrlValidator();
    String region = "";
    TrackFormat fmt = Utils.getFileTypeFromName(x);
    if (fmt.equals(TrackFormat.BAM)) {
        SamReader samReader;
        if (urlValidator.isValid(x)) {
            samReader = SamReaderFactory.makeDefault().open(SamInputResource.of(new URL(x)));
        } else {
            SamReaderFactory srf = SamReaderFactory.make();
            srf.validationStringency(ValidationStringency.SILENT);
            samReader = srf.open(new File(x));
        }
        // Default: Start from the first contig in dictionary
        region = samReader.getFileHeader().getSequence(0).getSequenceName();
        SAMRecordIterator iter = samReader.iterator();
        if (iter.hasNext()) {
            // If there are records in this BAM, init from first record
            SAMRecord rec = iter.next();
            region = rec.getContig() + ":" + rec.getAlignmentStart();
            samReader.close();
        }
        return region;
    } else if (fmt.equals(TrackFormat.BIGWIG) && !urlValidator.isValid(x)) {
        // Loading from URL is painfully slow so do not initialize from URL
        return initRegionFromBigWig(x);
    } else if (fmt.equals(TrackFormat.BIGBED) && !urlValidator.isValid(x)) {
        // Loading from URL is painfully slow so do not initialize from URL
        return initRegionFromBigBed(x);
    } else if (urlValidator.isValid(x) && (fmt.equals(TrackFormat.BIGWIG) || fmt.equals(TrackFormat.BIGBED))) {
        System.err.println("Refusing to initialize from URL");
        throw new InvalidGenomicCoordsException();
    } else if (fmt.equals(TrackFormat.TDF)) {
        Iterator<String> iter = TDFReader.getReader(x).getChromosomeNames().iterator();
        while (iter.hasNext()) {
            region = iter.next();
            if (!region.equals("All")) {
                return region;
            }
        }
        System.err.println("Cannot initialize from " + x);
        throw new RuntimeException();
    } else {
        // Input file appears to be a generic interval file. We expect chrom to be in column 1
        // VCF files are also included here since they are either gzip or plain ASCII.
        BufferedReader br;
        GZIPInputStream gzipStream;
        if (x.toLowerCase().endsWith(".gz") || x.toLowerCase().endsWith(".bgz")) {
            if (urlValidator.isValid(x)) {
                gzipStream = new GZIPInputStream(new URL(x).openStream());
            } else {
                InputStream fileStream = new FileInputStream(x);
                gzipStream = new GZIPInputStream(fileStream);
            }
            Reader decoder = new InputStreamReader(gzipStream, "UTF-8");
            br = new BufferedReader(decoder);
        } else {
            if (urlValidator.isValid(x)) {
                InputStream instream = new URL(x).openStream();
                Reader decoder = new InputStreamReader(instream, "UTF-8");
                br = new BufferedReader(decoder);
            } else {
                br = new BufferedReader(new FileReader(x));
            }
        }
        String line;
        while ((line = br.readLine()) != null) {
            line = line.trim();
            if (line.startsWith("#") || line.isEmpty() || line.startsWith("track ")) {
                continue;
            }
            if (fmt.equals(TrackFormat.VCF)) {
                region = line.split("\t")[0] + ":" + line.split("\t")[1];
            } else {
                IntervalFeature feature = new IntervalFeature(line, fmt, null);
                region = feature.getChrom() + ":" + feature.getFrom();
            }
            br.close();
            return region;
        }
        if (line == null) {
            // This means the input has no records
            region = "Undefined_contig";
            if (fmt.equals(TrackFormat.VCF)) {
                SAMSequenceDictionary seqdict = getVCFHeader(x).getSequenceDictionary();
                if (seqdict != null) {
                    Iterator<SAMSequenceRecord> iter = seqdict.getSequences().iterator();
                    if (iter.hasNext()) {
                        region = iter.next().getSequenceName();
                    }
                }
            }
            return region;
        }
    }
    System.err.println("Cannot initialize from " + x);
    throw new RuntimeException();
}