Examples of TabbedTextFileWithHeaderParser

picard.util.TabbedTextFileWithHeaderParser
Parse a tabbed text file in which columns are found by looking at a header line rather than by position. @author alecw@broadinstitute.org

Examples of picard.util.TabbedTextFileWithHeaderParser

    private static final String BARCODE_SEQUENCE_1_COLUMN = "barcode_sequence_1";
    private static final String BARCODE_NAME_COLUMN = "barcode_name";
    private static final String LIBRARY_NAME_COLUMN = "library_name";


    private void parseBarcodeFile(final ArrayList<String> messages) {
        final TabbedTextFileWithHeaderParser barcodesParser = new TabbedTextFileWithHeaderParser(BARCODE_FILE);
        final String sequenceColumn = barcodesParser.hasColumn(BARCODE_SEQUENCE_COLUMN)
                ? BARCODE_SEQUENCE_COLUMN : barcodesParser.hasColumn(BARCODE_SEQUENCE_1_COLUMN)
                ? BARCODE_SEQUENCE_1_COLUMN : null;
        if (sequenceColumn == null) {
            messages.add(BARCODE_FILE + " does not have " + BARCODE_SEQUENCE_COLUMN + " or " +
                    BARCODE_SEQUENCE_1_COLUMN + " column header");
            return;
        }
        final boolean hasBarcodeName = barcodesParser.hasColumn(BARCODE_NAME_COLUMN);
        final boolean hasLibraryName = barcodesParser.hasColumn(LIBRARY_NAME_COLUMN);
        final int numBarcodes = readStructure.barcodes.length();
        final Set<String> barcodes = new HashSet<String>();
        for (final TabbedTextFileWithHeaderParser.Row row : barcodesParser) {
            final String bcStrings[] = new String[numBarcodes];
            int barcodeNum = 1;
            for (final ReadDescriptor rd : readStructure.descriptors) {
                if (rd.type != ReadType.Barcode) continue;
                final String header = barcodeNum == 1 ? sequenceColumn : "barcode_sequence_" + String.valueOf(barcodeNum);
                bcStrings[barcodeNum - 1] = row.getField(header);
                barcodeNum++;
            }
            final String bcStr = IlluminaUtil.barcodeSeqsToString(bcStrings);
            if (barcodes.contains(bcStr)) {
                messages.add("Barcode " + bcStr + " specified more than once in " + BARCODE_FILE);
            }
            barcodes.add(bcStr);
            final String barcodeName = (hasBarcodeName ? row.getField(BARCODE_NAME_COLUMN) : "");
            final String libraryName = (hasLibraryName ? row.getField(LIBRARY_NAME_COLUMN) : "");
            final BarcodeMetric metric = new BarcodeMetric(barcodeName, libraryName, bcStr, bcStrings);
            barcodeToMetrics.put(StringUtil.join("", bcStrings), metric);
        }
        barcodesParser.close();
    }

View Full Code Here

Examples of picard.util.TabbedTextFileWithHeaderParser

            factory = new IlluminaDataProviderFactory(BASECALLS_DIR, LANE, readStructure, bclQualityEvaluationStrategy,
                    IlluminaDataType.PF, IlluminaDataType.Position);
        } else {
            // Grab expected barcode data from barcodeData.<LANE>
            IOUtil.assertFileIsReadable(INPUT);
            final TabbedTextFileWithHeaderParser barcodesParser = new TabbedTextFileWithHeaderParser(INPUT);
            for (final TabbedTextFileWithHeaderParser.Row row : barcodesParser) {
                final String barcodeName = row.getField(BARCODE_NAME_COLUMN);
                final StringBuilder barcode = new StringBuilder();
                for (int i = 1; i <= readStructure.barcodes.length(); i++) {
                    barcode.append(row.getField(BARCODE_SEQUENCE_COLUMN_NAME_STUB + i));

View Full Code Here

Examples of picard.util.TabbedTextFileWithHeaderParser

    /**
     * For each line in the LIBRARY_PARAMS file create a SamFileWriter and put it in the barcodeSamWriterMap map, where
     * the key to the map is the concatenation of all barcodes in order for the given line
     */
    private void populateWritersFromLibraryParams() {
        final TabbedTextFileWithHeaderParser libraryParamsParser = new TabbedTextFileWithHeaderParser(LIBRARY_PARAMS);


        final Set<String> expectedColumnLabels = CollectionUtil.makeSet("OUTPUT", "SAMPLE_ALIAS", "LIBRARY_NAME");
        final List<String> barcodeColumnLabels = new ArrayList<String>();
        if (readStructure.barcodes.length() == 1) {
            //For the single barcode read case, the barcode label name can either by BARCODE or BARCODE_1
            if (libraryParamsParser.hasColumn("BARCODE")) {
                barcodeColumnLabels.add("BARCODE");
            } else if (libraryParamsParser.hasColumn("BARCODE_1")) {
                barcodeColumnLabels.add("BARCODE_1");
            } else {
                throw new PicardException("LIBRARY_PARAMS(BARCODE_PARAMS) file " + LIBRARY_PARAMS + " does not have column BARCODE or BARCODE_1.");
            }
        } else {
            for (int i = 1; i <= readStructure.barcodes.length(); i++) {
                barcodeColumnLabels.add("BARCODE_" + i);
            }
        }


        expectedColumnLabels.addAll(barcodeColumnLabels);
        final Set<String> rgTagColumns = findAndFilterExpectedColumns(libraryParamsParser.columnLabels(), expectedColumnLabels);
        checkRgTagColumns(rgTagColumns);


        for (final TabbedTextFileWithHeaderParser.Row row : libraryParamsParser) {
            List<String> barcodeValues = null;


            if (barcodeColumnLabels.size() > 0) {
                barcodeValues = new ArrayList<String>();
                for (final String barcodeLabel : barcodeColumnLabels) {
                    barcodeValues.add(row.getField(barcodeLabel));
                }
            }


            final String key = (barcodeValues == null || barcodeValues.contains("N")) ? null : StringUtil.join("", barcodeValues);
            if (barcodeSamWriterMap.containsKey(key)) {    //This will catch the case of having more than 1 line in a non-barcoded LIBRARY_PARAMS file
                throw new PicardException("Row for barcode " + key + " appears more than once in LIBRARY_PARAMS or BARCODE_PARAMS file " +
                        LIBRARY_PARAMS);
            }


            final Map<String, String> samHeaderParams = buildSamHeaderParameters(barcodeValues);


            for (final String tagName : rgTagColumns) {
                samHeaderParams.put(tagName, row.getField(tagName));
            }


            final SAMFileWriterWrapper writer = buildSamFileWriter(new File(row.getField("OUTPUT")),
                    row.getField("SAMPLE_ALIAS"), row.getField("LIBRARY_NAME"), samHeaderParams);
            barcodeSamWriterMap.put(key, writer);
        }
        if (barcodeSamWriterMap.isEmpty()) {
            throw new PicardException("LIBRARY_PARAMS(BARCODE_PARAMS) file " + LIBRARY_PARAMS + " does have any data rows.");
        }
        libraryParamsParser.close();
    }

View Full Code Here

Examples of picard.util.TabbedTextFileWithHeaderParser


    OverlapDetector<Gene> load() {
        final OverlapDetector<Gene> overlapDetector = new OverlapDetector<Gene>(0, 0);


        final int expectedColumns = RefFlatColumns.values().length;
        final TabbedTextFileWithHeaderParser parser = new TabbedTextFileWithHeaderParser(refFlatFile, RefFlatColumnLabels);
        final Map<String, List<TabbedTextFileWithHeaderParser.Row>> refFlatLinesByGene =
                new HashMap<String, List<TabbedTextFileWithHeaderParser.Row>>();


        for (final TabbedTextFileWithHeaderParser.Row row : parser) {
            final int lineNumber = parser.getCurrentLineNumber(); // getCurrentLineNumber returns the number of the next line
            if (row.getFields().length != expectedColumns) {
                throw new AnnotationException("Wrong number of fields in refFlat file " + refFlatFile + " at line " +
                        lineNumber);
            }
            final String geneName = row.getField(RefFlatColumns.GENE_NAME.name());

View Full Code Here

Examples of picard.util.TabbedTextFileWithHeaderParser

    /**
     * For each line in the MULTIPLEX_PARAMS file create a FastqRecordsWriter and put it in the barcodeFastqWriterMap map,
     * where the key to the map is the concatenation of all barcodes in order for the given line.
     */
    private void populateWritersFromMultiplexParams() {
        final TabbedTextFileWithHeaderParser libraryParamsParser = new TabbedTextFileWithHeaderParser(MULTIPLEX_PARAMS);


        final Set<String> expectedColumnLabels = CollectionUtil.makeSet("OUTPUT_PREFIX");
        final List<String> barcodeColumnLabels = new ArrayList<String>();
        for (int i = 1; i <= readStructure.barcodes.length(); i++) {
            barcodeColumnLabels.add("BARCODE_" + i);
        }


        expectedColumnLabels.addAll(barcodeColumnLabels);
        assertExpectedColumns(libraryParamsParser.columnLabels(), expectedColumnLabels);


        for (final TabbedTextFileWithHeaderParser.Row row : libraryParamsParser) {
            List<String> barcodeValues = null;


            if (barcodeColumnLabels.size() > 0) {
                barcodeValues = new ArrayList<String>();
                for (final String barcodeLabel : barcodeColumnLabels) {
                    barcodeValues.add(row.getField(barcodeLabel));
                }
            }


            final String key = (barcodeValues == null || barcodeValues.contains("N")) ? null : StringUtil.join("", barcodeValues);
            if (barcodeFastqWriterMap.containsKey(key)) {    //This will catch the case of having more than 1 line in a non-barcoded MULTIPLEX_PARAMS file
                throw new PicardException("Row for barcode " + key + " appears more than once in MULTIPLEX_PARAMS file " +
                        MULTIPLEX_PARAMS);
            }


            final FastqRecordsWriter writer = buildWriter(new File(row.getField("OUTPUT_PREFIX")));
            barcodeFastqWriterMap.put(key, writer);
        }
        if (barcodeFastqWriterMap.isEmpty()) {
            throw new PicardException("MULTIPLEX_PARAMS file " + MULTIPLEX_PARAMS + " does have any data rows.");
        }
        libraryParamsParser.close();
    }

View Full Code Here

TOP

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.