v5.1.6

Updates and fixes in this release

Pisces

Somatic Calling

• Improve somatic accuracy by improving Q scores accuracy

Germline Calling

• Update for compatibility with diploid phasing
• Clean up overlapping variants into two separate alleles, as MNVs
• Accurate functional consequence through complex regions
• Location merging for not-co-located 1/2 GT
• Genotype reporting at loci inside deletions and MNVs reported at earlier loci

Filtering

• Allow application of RMxN filter to depend on variant frequency

Usability and Performance

• Remove depracated aliases for command line options and make input arguments case insensitive
• Update default value for Collapse parameter to true
• Remove GTModel parameter
• If supplied output folder doesn't exist, be kind to the user and create it
• Remove internal stitching in favor of separate read-stitching executable
• Turn off interval jumping during BAM reading

Bug Fixes

• Some references were getting a PASS when they failed Q-score but did not fail depth (rare case, when a deletion or MNV would eat up all the other coverage, so coverage was high, but ref support was low enough to flunk Q score)
• Fix bug in rounding when calculating coverage/variant frequency over a spanning stitched variant
• Fix germline location merging for indels

New software in this release

Stitcher

The Stitcher reconciles paired alignments in a supplied BAM file, outputting a "stitched" BAM where each valid overlapping pair is merged into a single read. This allows for more accurate, fragment-based representation of paired end data - both in terms of coverage and of base-calling/alignment confidence by addressing agreements and disagreements in overlapping regions of paired reads.

Details at: Stitcher-5.1.6-Design-Document

Scylla

Detects multiple nucleotide variants (MNVs) in a given sample and phases the variants in complex regions into sub populations. Allows phasing of variants across reads.

Details at: Scylla-5.1.6-Design-Document

VariantQualityRecalibration

Recalibrates the variant quality scores (Q scores) if the particular variants are over represented

Details at: VQR-5.1.6-Design-Document