Examples with ReadCoordinateComparator org.broadinstitute.hellbender.utils.read.ReadCoordinateComparator used on opensource projects

Example 6 with ReadCoordinateComparator

use of org.broadinstitute.hellbender.utils.read.ReadCoordinateComparator in project gatk by broadinstitute.

the class MarkDuplicatesSparkUtils method handleFragments.

private static List<GATKRead> handleFragments(Iterable<PairedEnds> pairedEnds, final MarkDuplicatesScoringStrategy scoringStrategy, final SAMFileHeader header) {
    List<GATKRead> reads = Lists.newArrayList();
    final Iterable<GATKRead> transform = Iterables.transform(pairedEnds, pair -> pair.first());
    Iterable<GATKRead> readsCopy = Iterables.transform(transform, GATKRead::copy);
    final Map<Boolean, List<GATKRead>> byPairing = Utils.stream(readsCopy).collect(Collectors.partitioningBy(read -> ReadUtils.readHasMappedMate(read)));
    // Note the we emit only fragments from this mapper.
    if (byPairing.get(true).isEmpty()) {
        // There are no paired reads, mark all but the highest scoring fragment as duplicate.
        Comparator<GATKRead> fragmentsComparator = Comparator.<GATKRead, Integer>comparing(read -> scoringStrategy.score(read)).reversed().thenComparing(new ReadCoordinateComparator(header));
        List<GATKRead> frags = byPairing.get(false).stream().sorted(fragmentsComparator).collect(Collectors.toList());
        if (!frags.isEmpty()) {
            //highest score - just emit
            reads.add(frags.get(0));
            for (final GATKRead record : Iterables.skip(frags, 1)) {
                //lower   scores - mark as dups and emit
                record.setIsDuplicate(true);
                reads.add(record);
            }
        }
    } else {
        // There are paired ends so we mark all fragments as duplicates.
        for (final GATKRead record : byPairing.get(false)) {
            record.setIsDuplicate(true);
            reads.add(record);
        }
    }
    return reads;
}

Example 7 with ReadCoordinateComparator

use of org.broadinstitute.hellbender.utils.read.ReadCoordinateComparator in project gatk by broadinstitute.

the class CompareDuplicatesSpark method getDupes.

/**
     * getDupes returns the metadata about how well the two sets of reads match.
     * @param f reads from the first bam
     * @param s reads from the second bam
     * @param header header (should be the same for both)
     * @return the type of the match, EQUAL, DIFFERENT_REPRESENTATIVE_READ, etc.
     */
static MatchType getDupes(Iterable<GATKRead> f, Iterable<GATKRead> s, SAMFileHeader header) {
    List<GATKRead> first = Lists.newArrayList(f);
    List<GATKRead> second = Lists.newArrayList(s);
    if (first.size() != second.size()) {
        return MatchType.SIZE_UNEQUAL;
    }
    int size = first.size();
    first.sort(new ReadCoordinateComparator(header));
    second.sort(new ReadCoordinateComparator(header));
    Set<GATKRead> firstDupes = Sets.newLinkedHashSet();
    Set<GATKRead> secondDupes = Sets.newLinkedHashSet();
    for (int i = 0; i < size; ++i) {
        GATKRead firstRead = first.get(i);
        GATKRead secondRead = second.get(i);
        if (!firstRead.getName().equals(secondRead.getName())) {
            return MatchType.READ_MISMATCH;
        }
        if (firstRead.isDuplicate()) {
            firstDupes.add(firstRead);
        }
        if (secondRead.isDuplicate()) {
            secondDupes.add(secondRead);
        }
    }
    if (firstDupes.size() != secondDupes.size()) {
        return MatchType.DIFF_NUM_DUPES;
    }
    if (!firstDupes.equals(secondDupes)) {
        return MatchType.DIFFERENT_REPRESENTATIVE_READ;
    }
    return MatchType.EQUAL;
}