Examples with DataLine org.broadinstitute.hellbender.utils.tsv.DataLine used on opensource projects

Example 1 with DataLine

use of org.broadinstitute.hellbender.utils.tsv.DataLine in project gatk by broadinstitute.

the class CombineReadCounts method readCountFileReader.

/**
     * Creates a read-count file reader given the input files and the expected target collection.
     * @param file the input file.
     * @param targets the expected targets in the input file.
     * @return never {@code null}.
     */
private TableReader<ReadCountRecord> readCountFileReader(final File file, final TargetCollection<Target> targets) {
    try {
        return new TableReader<ReadCountRecord>(file) {

            private boolean hasName;

            private boolean hasCoordinates;

            private int countColumnCount;

            private int[] countColumnIndexes;

            @Override
            public void processColumns(final TableColumnCollection columns) {
                hasCoordinates = columns.containsAll(TargetTableColumn.CONTIG.toString(), TargetTableColumn.START.toString(), TargetTableColumn.END.toString());
                hasName = columns.contains(TargetTableColumn.NAME.toString());
                if (!hasCoordinates && !hasName) {
                    throw formatException("header contain neither coordinates nor target name columns");
                }
                final List<String> countColumnNames = readCountColumnNames(columns);
                countColumnCount = countColumnNames.size();
                countColumnIndexes = new int[countColumnCount];
                for (int i = 0; i < countColumnCount; i++) {
                    countColumnIndexes[i] = columns.indexOf(countColumnNames.get(i));
                }
            }

            @Override
            protected ReadCountRecord createRecord(final DataLine dataLine) {
                final double[] counts = new double[countColumnCount];
                final Target target = createTarget(dataLine);
                for (int i = 0; i < counts.length; i++) {
                    counts[i] = dataLine.getDouble(countColumnIndexes[i]);
                }
                return new ReadCountRecord(target, counts);
            }

            /**
                 * Extracts the target object out of a data input line.
                 * @param dataLine the input data line.
                 * @return never {@code null}.
                 */
            private Target createTarget(final DataLine dataLine) {
                if (hasName) {
                    final String name = dataLine.get(TargetTableColumn.NAME);
                    final Target target = targets.target(name);
                    final SimpleInterval interval = createInterval(dataLine);
                    if (target == null) {
                        return new Target(name, createInterval(dataLine));
                    } else if (interval != null && !interval.equals(target.getInterval())) {
                        throw new UserException.BadInput(String.format("invalid target '%s' coordinates: expected %s but found %s", name, target.getInterval(), createInterval(dataLine)));
                    } else {
                        return target;
                    }
                } else {
                    // hasCoordinates must be true.
                    final SimpleInterval interval = createInterval(dataLine);
                    final Optional<Target> target = targets.targets(interval).stream().findAny();
                    if (!target.isPresent() || !target.get().getInterval().equals(interval)) {
                        throw formatException("target not found with coordinates " + interval);
                    }
                    return target.get();
                }
            }

            /**
                 * Extract the interval out of a data line.
                 * @param dataLine the input data line.
                 * @return {@code null} if the interval cannot be determined from the input file alone.
                 */
            private SimpleInterval createInterval(final DataLine dataLine) {
                if (hasCoordinates) {
                    return new SimpleInterval(dataLine.get(TargetTableColumn.CONTIG), dataLine.getInt(TargetTableColumn.START), dataLine.getInt(TargetTableColumn.END));
                } else {
                    return null;
                }
            }
        };
    } catch (final IOException ex) {
        throw new UserException.CouldNotReadInputFile(file, ex);
    }
}

Example 2 with DataLine

use of org.broadinstitute.hellbender.utils.tsv.DataLine in project gatk by broadinstitute.

the class ReadCountCollectionUtils method createReadCountRecordTableWriterWithoutIntervals.

private static TableWriter<ReadCountRecord> createReadCountRecordTableWriterWithoutIntervals(final Writer writer, List<String> columnNames) throws IOException {
    final TableColumnCollection columns = new TableColumnCollection(columnNames);
    return new TableWriter<ReadCountRecord>(writer, columns) {

        @Override
        protected void composeLine(final ReadCountRecord record, final DataLine dataLine) {
            dataLine.append(record.getTarget().getName());
            record.appendCountsTo(dataLine);
        }
    };
}

Example 3 with DataLine

use of org.broadinstitute.hellbender.utils.tsv.DataLine in project gatk-protected by broadinstitute.

the class ReadCountCollectionUtils method writerWithIntervals.

/**
     * Creates a new table writer that will output the target intervals.
     * @param writer where to output the table formatted content.
     * @param countColumnNames list of count column names.
     * @return never {@code null}.
     * @throws IOException if there is some low level IO problem creating the writer.
     * @throws IllegalArgumentException if {@code countColumnNames} is {@code null}, contains
     *  {@code null} or a non valid count column name (e.g. a reserved word).
     */
public static TableWriter<ReadCountRecord> writerWithIntervals(final Writer writer, final List<String> countColumnNames) throws IOException {
    final List<String> columnNames = new ArrayList<>();
    columnNames.add(TargetTableColumn.CONTIG.toString());
    columnNames.add(TargetTableColumn.START.toString());
    columnNames.add(TargetTableColumn.END.toString());
    columnNames.add(TargetTableColumn.NAME.toString());
    columnNames.addAll(Utils.nonNull(countColumnNames));
    final TableColumnCollection columns = new TableColumnCollection(columnNames);
    return new TableWriter<ReadCountRecord>(writer, columns) {

        @Override
        protected void composeLine(final ReadCountRecord record, final DataLine dataLine) {
            final SimpleInterval interval = record.getTarget().getInterval();
            if (interval == null) {
                throw new IllegalStateException("invalid combination of targets with and without intervals defined");
            }
            dataLine.append(interval.getContig()).append(interval.getStart()).append(interval.getEnd()).append(record.getTarget().getName());
            record.appendCountsTo(dataLine);
        }
    };
}

Example 4 with DataLine

use of org.broadinstitute.hellbender.utils.tsv.DataLine in project gatk-protected by broadinstitute.

the class ReadCountCollectionUtils method createReadCountRecordTableWriterWithoutIntervals.

private static TableWriter<ReadCountRecord> createReadCountRecordTableWriterWithoutIntervals(final Writer writer, List<String> columnNames) throws IOException {
    final TableColumnCollection columns = new TableColumnCollection(columnNames);
    return new TableWriter<ReadCountRecord>(writer, columns) {

        @Override
        protected void composeLine(final ReadCountRecord record, final DataLine dataLine) {
            dataLine.append(record.getTarget().getName());
            record.appendCountsTo(dataLine);
        }
    };
}

Example 5 with DataLine

use of org.broadinstitute.hellbender.utils.tsv.DataLine in project gatk-protected by broadinstitute.

the class XHMMSegmentCallerBaseIntegrationTest method writeChainInTempFile.

public static File writeChainInTempFile(final XHMMData chain) {
    final File result = createTempFile("chain-", ".tab");
    //final File result = new File("/tmp/input");
    final List<String> sampleNames = IntStream.range(0, chain.data.size()).mapToObj(a -> "SAMPLE_" + a).collect(Collectors.toList());
    final List<String> columnNames = new ArrayList<>(sampleNames.size() + 4);
    columnNames.addAll(Arrays.asList(TargetTableColumn.CONTIG.toString(), TargetTableColumn.START.toString(), TargetTableColumn.END.toString(), TargetTableColumn.NAME.toString()));
    columnNames.addAll(sampleNames);
    try (final TableWriter<Integer> writer = new TableWriter<Integer>(result, new TableColumnCollection(columnNames)) {

        @Override
        protected void composeLine(final Integer record, final DataLine dataLine) {
            dataLine.append(chain.targets.get(record).getContig()).append(chain.targets.get(record).getStart()).append(chain.targets.get(record).getEnd()).append(chain.targets.get(record).getName());
            for (final List<Double> sampleData : chain.data) {
                dataLine.append(sampleData.get(record));
            }
        }
    }) {
        writer.writeAllRecords(IntStream.range(0, chain.targets.size()).boxed().collect(Collectors.toList()));
    } catch (final IOException ex) {
        throw new UncheckedIOException(ex);
    }
    return result;
}