diff --git a/src/main/java/org/broadinstitute/hellbender/tools/walkers/groundtruth/GroundTruthScorer.java b/src/main/java/org/broadinstitute/hellbender/tools/walkers/groundtruth/GroundTruthScorer.java index ff44ac5ca5d..f3191fbaadb 100644 --- a/src/main/java/org/broadinstitute/hellbender/tools/walkers/groundtruth/GroundTruthScorer.java +++ b/src/main/java/org/broadinstitute/hellbender/tools/walkers/groundtruth/GroundTruthScorer.java @@ -37,6 +37,99 @@ import java.util.*; import java.util.zip.GZIPOutputStream; +/** + * Converts Ultima reads into flow-based annotation, and provides some general statistics regarding + * quality and errors relative to the reference. Ultima data is flow-based, and thus the original computed + * quality refers to each flow, rather than each base. In the Ultima cram/bam, there is a per-base representation + * of the original flow qualities, where the original quality is distributed along each flow (homopolymer). + * In order to reconstitute the original flow information, the tool incorporates the information encoded + * in the Ultima cram/bam, and outputs both the read in flow space, as well as a conversion of the aligned + * reference portion into flow space, and an alignment score. + * + *

Input

+ * + * + *

Output

+ * + * + *

CSV Output Description

+ * csv with the read representation in flow space. The csv includes the following columns: + *
  • ReadName
    • + *
  • ReadKey : The signal of the read at each flow according to the flow order
    • + *
  • ReadIsReversed : Whether the read is reversed in the alignment
    • + *
  • ReadMQ : The mapping quality of the read
    • + *
  • ReadRQ : The read rq value
    • + *
  • GroundTruthKey : The aligned reference section, translated into per-flow signals
    • + *
  • ReadSequence
    • + *
  • Score : A flow-based alignment score. Since the alignment is per-flow, in the case that there’s a cycle skip, the read and reference flow signals will not be aligned, and therefore the score will be inaccurate.
    • + *
  • NormalizedScore: A flow-based normalized alignment score
    • + *
  • ErrorProbability : The error of each flow (corresponds to the signals in ReadKey)
    • + *
  • ReadKeyLength
    • + *
  • GroundTruthKeyLength
    • + *
  • CycleSkipStatus : One of NS (Non Skip), PCS (Possible Cycle Skip), or CS (Cycle Skip)
    • + *
  • Cigar
    • + *
  • LowestQBaseTP
    • + * + *

    GATK Report Description

    + * In the quality report (optional), the following tables are included: + * + *
  • qualReport:error rate per qual : The error rate for each quality. Columns:
    • + * + * + *
  • qual_hmerReport:error rate per qual by hmer. The error rate for each quality and hmer combination. Columns:
    • + * + * + *
  • qual_hmer_deviation_base_Report:error rate per qual by hmer and deviation. The count of errors for each qual, hmer, deviation and base
    • + * + * + *
  • Phred/qual statistics per flow position report. Various statistics for each flow position in relationship to the found quality value. Columns:
    • + * + * + *

    Usage examples

    + * 
    + * gatk GroundTruthScorer \
+ *   -I input.bam \
+ *   -R reference.fasta.gz
+ *   -L chr20 \
+ *   --output-csv output.csv \
+ *   --report-file report.txt \
+ *   --omit-zeros-from-report \ (optional)
+ *   --features-file dbsnp.chr9.vcf.gz \ (optional)
+ *   --genome-prior genome_prior.csv (optional)
+ *
    + * + * {@GATK.walkertype ReadWalker} + */ + @CommandLineProgramProperties( summary = "Ground Truth Scorer", oneLineSummary = "Score reads against a reference/ground truth",