Examples with Barcode cz1.gbs.core.Barcode used on opensource projects

Example 1 with Barcode

use of cz1.gbs.core.Barcode in project polyGembler by c-zhou.

the class ParseBarcodeRead method setupBarcodeFiles.

/**
 * Reads in an Illumina key file, creates a linear array of {@link Barcode} objects
 * representing the barcodes in the key file, then creates a hash map containing
 * indices from the linear array indexed by sequence.  The names of barcode objects
 * follow the pattern samplename:flowcell:lane:well, since sample names alone are not unique.
 *
 * @param keyFile Illumina key file.
 * @param flowcell Only barcodes from this flowcell will be added to the array.
 * @param lane Only barcodes from this lane will be added to the array.
 * @return Number of barcodes in the array.
 */
private int setupBarcodeFiles(File keyFile, String flowcell, String lane) {
    try {
        BufferedReader br = new BufferedReader(new FileReader(keyFile), 65536);
        ArrayList<Barcode> theBarcodesArrayList = new ArrayList<Barcode>();
        String temp;
        int k = 0;
        while (((temp = br.readLine()) != null)) {
            // split by whitespace
            String[] s = temp.split("\\t");
            Barcode theBC = null;
            if (s[0].equals(flowcell) && s[1].equals(lane)) {
                String well = (s[6].length() < 2) ? (s[5] + '0' + s[6]) : s[5] + s[6];
                if (s.length < 8 || s[7] == null || s[7].equals("")) {
                    // use the plate and well
                    theBC = new Barcode(s[2], initialCutSiteRemnant, s[3] + ":" + s[0] + ":" + s[1] + ":" + s[4] + ":" + well, flowcell, lane, k++);
                } else {
                    // use the "libraryPlateWellID" or whatever is in column H of the key file, IF it is an integer
                    try {
                        int libPrepID = Integer.parseInt(s[7]);
                        theBC = new Barcode(s[2], initialCutSiteRemnant, s[3] + ":" + s[0] + ":" + s[1] + ":" + libPrepID, flowcell, lane, k++);
                    } catch (NumberFormatException nfe) {
                        theBC = new Barcode(s[2], initialCutSiteRemnant, s[3] + ":" + s[0] + ":" + s[1] + ":" + s[4] + ":" + well, flowcell, lane, k++);
                    }
                }
                theBarcodesArrayList.add(theBC);
                System.out.println(theBC.getBarcodeString() + " " + theBC.getTaxaName());
            }
        }
        br.close();
        theBarcodes = new Barcode[theBarcodesArrayList.size()];
        theBarcodesArrayList.toArray(theBarcodes);
        Arrays.sort(theBarcodes);
        int nBL = theBarcodes[0].getBarOverLong().length;
        quickBarcodeList = new long[theBarcodes.length * nBL];
        quickMap = new HashMap<Long, Integer>();
        for (int i = 0; i < theBarcodes.length; i++) {
            for (int j = 0; j < nBL; j++) {
                quickBarcodeList[i * nBL + j] = theBarcodes[i].getBarOverLong()[j];
                quickMap.put(theBarcodes[i].getBarOverLong()[j], i);
            }
        }
        Arrays.sort(quickBarcodeList);
    } catch (Exception e) {
        System.out.println("Error with setupBarcodeFiles: " + e);
    }
    return theBarcodes.length;
}

Example 2 with Barcode

use of cz1.gbs.core.Barcode in project polyGembler by c-zhou.

the class ParseBarcodeRead method parseReadIntoTagAndTaxa.

/**
 * The barcode libraries used for this study can include two types of
 * extraneous sequence at the end of reads. The first are chimeras created
 * with the free ends. These will recreate the restriction site. The second
 * are short regions (less than 64bp), so that will they will contain a
 * portion of site and the universal adapter. This finds the first of site
 * in likelyReadEnd, keeps the restriction site overhang and then sets
 * everything to polyA afterwards
 *
 * @param seq An unprocessed tag sequence.
 * @param maxLength The maximum number of bp in the processed sequence.
 * @return returnValue A ReadBarcodeResult object containing the unprocessed
 * tag, Cut site position, Processed tag, and Poly-A padded tag.
 */
public ReadBarcodeResult parseReadIntoTagAndTaxa(String seqS, String qualS, int minQual) {
    if (seqS == null)
        return null;
    if (minQual != 0 && seqS != null) {
        try {
            final char[] seqC = (char[]) field.get(seqS), qualC = (char[]) field.get(qualS);
            int len = seqC.length;
            for (int i = 0; i < len; i++) if (qualC[i] - 33 < minQual)
                seqC[i] = 'N';
            seqS = String.valueOf(seqC);
        } catch (IllegalArgumentException | IllegalAccessException e) {
            // TODO Auto-generated catch block
            e.printStackTrace();
        }
    }
    Barcode bestBarcode = findBestBarcode(seqS, maximumMismatchInBarcodeAndOverhang);
    if (bestBarcode == null) {
        // overhang missing so skip
        return null;
    }
    seqS = StringUtils.strip(seqS.substring(bestBarcode.getBarLength()), "N");
    if (seqS.length() < 32)
        return null;
    int cutSitePosition = seqS.length();
    for (String end : likelyReadEnd) {
        int w = seqS.indexOf(end);
        if (w >= 0 && w < cutSitePosition)
            cutSitePosition = w;
    }
    String processedSeqS = StringUtils.strip(seqS.substring(0, cutSitePosition), "N");
    if (processedSeqS.length() >= 32)
        return new ReadBarcodeResult(BaseEncoder.getBitSetFromSeq(processedSeqS), bestBarcode.getTaxaId());
    return null;
}

Example 3 with Barcode

use of cz1.gbs.core.Barcode in project polyGembler by c-zhou.

the class FastqToTagSequence method run.

@Override
public void run() {
    String[] countFileNames = null;
    File inputDirectory = new File(this.myInputDirName);
    File[] fastqFiles = inputDirectory.listFiles(new FilenameFilter() {

        @Override
        public boolean accept(File dir, String name) {
            return name.matches("(?i).*\\.fq$|.*\\.fq\\.gz$|.*\\.fastq$|.*_fastq\\.txt$|.*_fastq\\.gz$|.*_fastq\\.txt\\.gz$|.*_sequence\\.txt$|.*_sequence\\.txt\\.gz$");
        // (?i) denotes case insensitive;                 \\. denotes escape . so it doesn't mean 'any char' & escape the backslash
        }
    });
    if (fastqFiles.length == 0 || fastqFiles == null) {
        myLogger.warn("Couldn't find any files that end with \".fq\", \".fq.gz\", \".fastq\", \"_fastq.txt\", \"_fastq.gz\", \"_fastq.txt.gz\", \"_sequence.txt\", or \"_sequence.txt.gz\" in the supplied directory.");
        return;
    } else {
        myLogger.info("Using the following FASTQ files:");
        countFileNames = new String[fastqFiles.length];
        for (int i = 0; i < fastqFiles.length; i++) {
            countFileNames[i] = fastqFiles[i].getName().replaceAll("(?i)\\.fq$|\\.fq\\.gz$|\\.fastq$|_fastq\\.txt$|_fastq\\.gz$|_fastq\\.txt\\.gz$|_sequence\\.txt$|_sequence\\.txt\\.gz$", "");
            // \\. escape . so it doesn't mean 'any char' & escape the backslash
            myLogger.info(fastqFiles[i].getAbsolutePath());
        }
    }
    for (int laneNum = 0; laneNum < fastqFiles.length; laneNum++) {
        // for (int laneNum = 2; laneNum < 3; laneNum++) {
        if (new File(myOutputDir + System.getProperty("file.separator") + countFileNames[laneNum] + ".cnt.gz").exists()) {
            myLogger.info("Fastq file " + fastqFiles[laneNum] + " skipped.");
            continue;
        }
        File outputFile = new File(this.myOutputDir + File.separator + countFileNames[laneNum]);
        if (outputFile.isFile()) {
            myLogger.warn("An output file " + countFileNames[laneNum] + "\n" + " already exists in the output directory for file " + fastqFiles[laneNum] + ".  Skipping.");
            continue;
        }
        myLogger.info("Reading FASTQ file: " + fastqFiles[laneNum]);
        String[] filenameField = fastqFiles[laneNum].getName().split("_");
        thePBR = new ParseBarcodeRead[this.myEnzyme.length];
        for (int i = 0; i < this.myEnzyme.length; i++) {
            if (filenameField.length == 3) {
                thePBR[i] = new ParseBarcodeRead(this.myKeyfile, this.myEnzyme[i], filenameField[0], filenameField[1]);
            } else if (filenameField.length == 4) {
                thePBR[i] = new ParseBarcodeRead(this.myKeyfile, this.myEnzyme[i], filenameField[0], filenameField[2]);
            } else // B08AAABXX_s_1_sequence.txt.gz
            if (filenameField.length == 5) {
                thePBR[i] = new ParseBarcodeRead(this.myKeyfile, this.myEnzyme[i], filenameField[1], filenameField[3]);
            } else {
                myLogger.error("Error in parsing file name: " + fastqFiles[laneNum]);
                myLogger.error("   The filename does not contain either 3, 4, or 5 underscore-delimited values.");
                myLogger.error("   Expect: flowcell_lane_fastq.txt.gz OR flowcell_s_lane_fastq.txt.gz OR code_flowcell_s_lane_fastq.txt.gz");
                continue;
            }
        }
        taxa = thePBR[0].getSortedTaxaNames();
        n = taxa.length;
        os = countFileNames[laneNum];
        volume = 0;
        myLogger.info("Total barcodes found in lane:" + thePBR[0].getBarCodeCount());
        if (thePBR[0].getBarCodeCount() == 0) {
            myLogger.warn("No barcodes found.  Skipping this flowcell lane.");
            continue;
        }
        String[] taxaNames = new String[thePBR[0].getBarCodeCount()];
        for (int i = 0; i < taxaNames.length; i++) {
            taxaNames[i] = thePBR[0].getTheBarcodes(i).getTaxaName();
        }
        long start = System.currentTimeMillis();
        this.initial_thread_pool();
        try {
            BufferedReader br = Utils.getBufferedReader(fastqFiles[laneNum], 65536);
            int block = 10000;
            String[][] Qs = new String[block][2];
            int k = 0;
            allReads = 0;
            goodBarcodedReads = 0;
            tags = 0;
            String temp = br.readLine();
            while (temp != null) {
                try {
                    Qs[k][0] = br.readLine();
                    br.readLine();
                    Qs[k][1] = br.readLine();
                } catch (NullPointerException e) {
                    myLogger.error("Unable to correctly parse the sequence from fastq file.  " + "Your fastq file may have been corrupted.");
                    System.exit(1);
                }
                k++;
                temp = br.readLine();
                if (k == block || temp == null) {
                    if (usedMemory() / maxMemory() > load) {
                        this.waitFor();
                        writeHardDisk();
                        this.initial_thread_pool();
                    }
                    executor.submit(new Runnable() {

                        private String[][] fastq;

                        @Override
                        public void run() {
                            // TODO Auto-generated method stub
                            try {
                                final Map<BitSet, short[]> block_tagCounts = new HashMap<BitSet, short[]>();
                                int block_allReads = 0, block_goodBarcodedReads = 0;
                                ReadBarcodeResult rr = null;
                                BitSet key;
                                for (int i = 0; i < fastq.length; i++) {
                                    if (fastq[i][0] == null)
                                        break;
                                    // synchronized(lock) {
                                    // allReads++;
                                    // }
                                    block_allReads++;
                                    outerloop: for (int j = 0; j < myLeadingTrim.length; j++) {
                                        for (int k = 0; k < thePBR.length; k++) {
                                            rr = thePBR[k].parseReadIntoTagAndTaxa(fastq[i][0].substring(myLeadingTrim[j]), fastq[i][1].substring(myLeadingTrim[j]), myMinQualS);
                                            if (rr != null)
                                                break outerloop;
                                        }
                                    }
                                    if (rr != null) {
                                        key = rr.read;
                                        /**
                                         *											synchronized(lock) {
                                         *												goodBarcodedReads++;
                                         *												if (allReads % 1000000 == 0) {
                                         *													myLogger.info("Total Reads:" + allReads +
                                         *															" Reads with barcode and cut site overhang:" +
                                         *															goodBarcodedReads);
                                         *												}
                                         *												if(!tagCounts.containsKey(key)) {
                                         *													tagCounts.put(key, new short[n]);
                                         *													tags++;
                                         *												}
                                         *												tagCounts.get(key)[rr.taxonId]++;
                                         *											}
                                         */
                                        block_goodBarcodedReads++;
                                        if (!block_tagCounts.containsKey(key)) {
                                            block_tagCounts.put(key, new short[n]);
                                        }
                                        block_tagCounts.get(key)[rr.taxonId]++;
                                    }
                                }
                                synchronized (lock) {
                                    allReads += block_allReads;
                                    goodBarcodedReads += block_goodBarcodedReads;
                                    // goodBarcodedReads);
                                    for (BitSet bs : block_tagCounts.keySet()) {
                                        if (!tagCounts.containsKey(bs)) {
                                            tags++;
                                            tagCounts.put(bs, block_tagCounts.get(bs));
                                        } else {
                                            short[] copy = tagCounts.get(bs);
                                            short[] block_copy = block_tagCounts.get(bs);
                                            for (int i = 0; i < n; i++) copy[i] += block_copy[i];
                                        }
                                    }
                                }
                            } catch (Exception e) {
                                Thread t = Thread.currentThread();
                                t.getUncaughtExceptionHandler().uncaughtException(t, e);
                                e.printStackTrace();
                                executor.shutdown();
                                System.exit(1);
                            }
                        }

                        public Runnable init(String[][] fastq) {
                            this.fastq = fastq;
                            return (this);
                        }
                    }.init(Qs));
                    k = 0;
                    Qs = new String[block][2];
                }
            }
            if (!tagCounts.isEmpty()) {
                this.waitFor();
                writeHardDisk();
            }
            br.close();
            myLogger.info("Total number of reads in lane=" + allReads);
            myLogger.info("Total number of good barcoded reads=" + goodBarcodedReads);
            myLogger.info("Total number of tags=" + tags);
            myLogger.info("Process took " + (System.currentTimeMillis() - start) / 1000 + " seconds.");
        } catch (Exception e) {
            e.printStackTrace();
        }
        myLogger.info("Finished reading " + (laneNum + 1) + " of " + fastqFiles.length + " sequence files.");
        tagCounts.clear();
    }
}

Example 4 with Barcode

use of cz1.gbs.core.Barcode in project polyGembler by c-zhou.

the class GBSpileup method setParameters.

@Override
public void setParameters(String[] args) {
    // TODO Auto-generated method stub
    if (args.length == 0) {
        printUsage();
        throw new IllegalArgumentException("\n\nPlease use the above arguments/options.\n\n");
    }
    if (myArgsEngine == null) {
        myArgsEngine = new ArgsEngine();
        myArgsEngine.add("-i", "--input-fastq", true);
        myArgsEngine.add("-k", "--key-file", true);
        myArgsEngine.add("-e", "--enzyme", true);
        myArgsEngine.add("-q", "--min-qualS", true);
        myArgsEngine.add("-p", "--ploidy", true);
        myArgsEngine.add("-t", "--threads", true);
        myArgsEngine.add("-T", "--trim-leading", true);
        myArgsEngine.add("-b", "--unassgined-reads", true);
        myArgsEngine.add("-f", "--reference", true);
        myArgsEngine.add("-z", "--skip-freebayes", false);
        myArgsEngine.add("-x", "--max-coverage", true);
        myArgsEngine.add("-o", "--prefix", true);
        myArgsEngine.parse(args);
    }
    if (myArgsEngine.getBoolean("-i")) {
        myInputDirName = myArgsEngine.getString("-i");
    } else {
        printUsage();
        throw new IllegalArgumentException("Please specify the location of your FASTQ files.");
    }
    if (myArgsEngine.getBoolean("-k")) {
        myKeyfile = myArgsEngine.getString("-k");
    } else {
        printUsage();
        throw new IllegalArgumentException("Please specify a barcode key file.");
    }
    if (myArgsEngine.getBoolean("-z")) {
        this.skipFB = true;
    } else {
        this.require("freebayes");
    }
    if (myArgsEngine.getBoolean("-f")) {
        myReference = myArgsEngine.getString("-f");
        if (!new File(myReference + ".amb").exists() || !new File(myReference + ".ann").exists() || !new File(myReference + ".bwt").exists() || !new File(myReference + ".pac").exists() || !new File(myReference + ".sa").exists()) {
            String index = "bwa index -p " + myReference + " -a bwtsw " + myReference;
            this.consume(this.bash(index));
        }
    } else {
        printUsage();
        throw new IllegalArgumentException("Please specify the reference.");
    }
    if (myArgsEngine.getBoolean("-e")) {
        myEnzyme = myArgsEngine.getString("-e").split("-");
    } else {
        myLogger.warn("No enzyme specified.  Using enzyme listed in key file.");
        try {
            BufferedReader br = Utils.getBufferedReader(myKeyfile);
            String[] s = br.readLine().split("\\s+");
            int k = -1;
            for (int i = 0; i < s.length; i++) if (s[i].toLowerCase().equals("enzyme"))
                k = i;
            if (k < 0)
                throw new IllegalArgumentException("No enzyme found in the key file. " + "Please specify the enzyme with -e option.\n\n");
            s = br.readLine().split("\\s+");
            myEnzyme = s[k].split("-");
            br.close();
        } catch (IOException e) {
            // TODO Auto-generated catch block
            e.printStackTrace();
        }
    }
    if (myArgsEngine.getBoolean("-q")) {
        myMinQualS = Integer.parseInt(myArgsEngine.getString("-q"));
    }
    if (myArgsEngine.getBoolean("-p")) {
        myPloidy = Integer.parseInt(myArgsEngine.getString("-p"));
    }
    if (myArgsEngine.getBoolean("-t")) {
        THREADS = Integer.parseInt(myArgsEngine.getString("-t"));
    }
    if (myArgsEngine.getBoolean("-x")) {
        maxCov = Long.parseLong(myArgsEngine.getString("-x"));
    }
    if (myArgsEngine.getBoolean("-T")) {
        int leading = Integer.parseInt(myArgsEngine.getString("-T"));
        if (leading > 0) {
            List<Integer> leadings = new ArrayList<Integer>();
            leadings.add(leading);
            for (int i = 1; i < 4; i++) {
                if (leading - i >= 0)
                    leadings.add(leading - i);
                leadings.add(leading + i);
            }
            myLeadingTrim = new int[leadings.size()];
            for (int i = 0; i < myLeadingTrim.length; i++) myLeadingTrim[i] = leadings.get(i);
        }
    }
    if (myArgsEngine.getBoolean("-o")) {
        myOutputDir = myArgsEngine.getString("-o");
    }
    File out = new File(myOutputDir);
    if (out.exists() && out.isDirectory()) {
        myLogger.warn("Output directory " + myOutputDir + " exsits. " + "We strongly recommend a new location.");
    }
}

Example 5 with Barcode

use of cz1.gbs.core.Barcode in project polyGembler by c-zhou.

the class TenXSamtools method setParameters.

@Override
public void setParameters(String[] args) {
    // TODO Auto-generated method stub
    if (args.length == 0) {
        printUsage();
        throw new IllegalArgumentException("\n\nPlease use the above arguments/options.\n\n");
    }
    switch(args[0].toUpperCase()) {
        case "SORT":
            this.task = Task.sort;
            break;
        default:
            printUsage();
            throw new IllegalArgumentException("\n\nPlease use the above arguments/options.\n\n");
    }
    String[] args2 = new String[args.length - 1];
    System.arraycopy(args, 1, args2, 0, args2.length);
    switch(this.task) {
        case sort:
            if (myArgsEngine == null) {
                myArgsEngine = new ArgsEngine();
                myArgsEngine.add("-i", "--in-bam", true);
                myArgsEngine.add("-b", "--barcode", false);
                myArgsEngine.add("-n", "--name", false);
                myArgsEngine.add("-s", "--batch-size", true);
                myArgsEngine.add("-t", "--threads", true);
                myArgsEngine.add("-o", "--out-bam", true);
                myArgsEngine.parse(args2);
            }
            if (myArgsEngine.getBoolean("-i")) {
                this.bam_in = myArgsEngine.getString("-i");
            } else {
                printUsage();
                throw new IllegalArgumentException("Please specify the input BAM file.");
            }
            if (myArgsEngine.getBoolean("-b") && myArgsEngine.getBoolean("-n")) {
                printUsage();
                throw new IllegalArgumentException("Options -b and -n are exculsive!!!");
            }
            if (myArgsEngine.getBoolean("-b")) {
                this.sort_order = Order.barcode;
                comprator = new Comparator<SAMRecord>() {

                    @Override
                    public int compare(SAMRecord record1, SAMRecord record2) {
                        // TODO Auto-generated method stub
                        if (record1 == null && record2 == null)
                            return 0;
                        if (record1 == null)
                            return 1;
                        if (record2 == null)
                            return -1;
                        String bx1 = record1.getStringAttribute("BX"), bx2 = record2.getStringAttribute("BX");
                        if (bx1 == null && bx2 == null)
                            return compareSAMRecord(record1, record2);
                        // none barcode record to the end
                        if (bx1 == null)
                            return 1;
                        if (bx2 == null)
                            return -1;
                        int diff = bx1.compareTo(bx2);
                        return diff == 0 ? compareSAMRecord(record1, record2) : diff;
                    }
                };
            }
            if (myArgsEngine.getBoolean("-n")) {
                this.sort_order = Order.queryname;
                comprator = new Comparator<SAMRecord>() {

                    @Override
                    public int compare(SAMRecord record1, SAMRecord record2) {
                        // TODO Auto-generated method stub
                        if (record1 == null && record2 == null)
                            return 0;
                        if (record1 == null)
                            return 1;
                        if (record2 == null)
                            return -1;
                        if (record1.getReadName().compareTo(record2.getReadName()) == 0)
                            return compareSAMRecord(record1, record2);
                        return record1.getReadName().compareTo(record2.getReadName());
                    }
                };
            }
            if (myArgsEngine.getBoolean("-s")) {
                this.batch_size = Integer.parseInt(myArgsEngine.getString("-s"));
            }
            if (myArgsEngine.getBoolean("-t")) {
                this.THREADS = Integer.parseInt(myArgsEngine.getString("-t"));
            }
            if (myArgsEngine.getBoolean("-o")) {
                this.bam_out = myArgsEngine.getString("-o");
            } else {
                printUsage();
                throw new IllegalArgumentException("Please specify the output BAM file.");
            }
            break;
        default:
            throw new RuntimeException("!!!");
    }
}