Examples with ReadTransformerSparkifier org.broadinstitute.hellbender.tools.spark.utils.ReadTransformerSparkifier used on opensource projects

Example 1 with ReadTransformerSparkifier

use of org.broadinstitute.hellbender.tools.spark.utils.ReadTransformerSparkifier in project gatk by broadinstitute.

the class PathSeqFilterSpark method runTool.

@Override
protected void runTool(final JavaSparkContext ctx) {
    final JavaRDD<GATKRead> reads = getReads();
    //Filter secondary/supplementary reads and reads that fail the vendor quality check
    final JavaRDD<GATKRead> primaryReads = reads.filter(read -> !(read.isSecondaryAlignment() || read.failsVendorQualityCheck() || read.isSupplementaryAlignment()));
    logger.info("Loaded " + reads.count() + " reads");
    //Mark and filter optical duplicates
    final OpticalDuplicateFinder finder = new OpticalDuplicateFinder(opticalDuplicatesArgumentCollection.READ_NAME_REGEX, opticalDuplicatesArgumentCollection.OPTICAL_DUPLICATE_PIXEL_DISTANCE, null);
    final JavaRDD<GATKRead> markedReads = MarkDuplicatesSpark.mark(primaryReads, getHeaderForReads(), MarkDuplicatesScoringStrategy.SUM_OF_BASE_QUALITIES, finder, getRecommendedNumReducers());
    final JavaRDD<GATKRead> markedFilteredReads = markedReads.filter(new ReadFilterSparkifier(new MarkedOpticalDuplicateReadFilter()));
    logger.info("Reads remaining after de-duplication: " + markedFilteredReads.count());
    //Apply DUST masking
    final JavaRDD<GATKRead> readsDUSTMasked = markedFilteredReads.map(new ReadTransformerSparkifier(new DUSTReadTransformer(DUST_MASK, DUST_W, DUST_T)));
    //Apply base quality hard clipping
    final JavaRDD<GATKRead> readsClipped = readsDUSTMasked.map(new ReadTransformerSparkifier(new BaseQualityClipReadTransformer(READ_TRIM_THRESH)));
    //Filter reads with less than MIN_READ_LENGTH bases
    final JavaRDD<GATKRead> readsLengthFiltered = readsClipped.filter(new ReadFilterSparkifier(new ReadLengthReadFilter(MIN_READ_LENGTH, Integer.MAX_VALUE)));
    logger.info("Reads remaining after clipping: " + readsLengthFiltered.count());
    //Change low-quality bases to 'N'
    final JavaRDD<GATKRead> readsBQFiltered = readsLengthFiltered.map(new ReadTransformerSparkifier(new BaseQualityReadTransformer(QUAL_PHRED_THRESH)));
    //Filter reads with too many 'N's
    final JavaRDD<GATKRead> readsAmbigFiltered = readsBQFiltered.filter(new ReadFilterSparkifier(new AmbiguousBaseReadFilter(FRAC_N_THRESHOLD)));
    logger.info("Reads remaining after ambiguous base filtering: " + readsAmbigFiltered.count());
    //Load Kmer hopscotch set and filter reads containing > 0 matching kmers
    final JavaRDD<GATKRead> readsKmerFiltered = doKmerFiltering(ctx, readsAmbigFiltered);
    logger.info("Reads remaining after kmer filtering: " + readsKmerFiltered.count());
    //Filter unpaired reads
    final JavaRDD<GATKRead> readsFilteredPaired = retainPairs(readsKmerFiltered);
    logger.info("Reads remaining after unpaired filtering: " + readsFilteredPaired.count());
    //BWA filtering against user-specified host organism reference
    header = getHeaderForReads();
    final JavaRDD<GATKRead> readsAligned = doHostBWA(ctx, header, readsFilteredPaired);
    //Get unmapped reads (note these always come in pairs)
    //TODO: retain read pairs by alignment score instead of flags
    final JavaRDD<GATKRead> readsNonHost = readsAligned.filter(read -> read.isUnmapped() && read.mateIsUnmapped());
    logger.info("Reads remaining after BWA filtering: " + readsFilteredPaired.count());
    //TODO: repeat BWA with seed size 11
    //Write output
    header.setSortOrder(SAMFileHeader.SortOrder.queryname);
    try {
        ReadsSparkSink.writeReads(ctx, OUTPUT_PATH, null, readsNonHost, header, shardedOutput ? ReadsWriteFormat.SHARDED : ReadsWriteFormat.SINGLE);
    } catch (final IOException e) {
        throw new GATKException("Unable to write bam", e);
    }
}