Skip to content

2.11.0

Compare
Choose a tag to compare
@d-cameron d-cameron released this 17 Mar 04:33
· 174 commits to master since this release

This release contains a new error correction step that improves assembly runtime, overhauls the VIRUSBreakend viral database, and bug fixes.

VIRUSBreakend databases need to be rebuilt.

  • Changed minimum clip length threshold from 50bp back to 5bp
    • Versions 2.10.x have lower sensitivity due to this regression. Rerunning with 2.11.0 is recommended
    • This regression primarily impacted single breakend sensitivity (since as split reads will generally have at least one side with 50bp clipped)
  • Improved assembly
    • Now performing local error correction of reads prior to assembly.
      • Improves assembly runtime and memory usage
    • Complex assembly graph regions are now downsampled instead of being excluded from assembly
    • Additional assembly performance optimisations
  • VIRUSBreakend
    • Extended database to include all NCBI viral neighbour sequences.
    • Deuplicating viral references with same taxid based on total kmer hits to each viral reference
    • Added LOW_MAPQ filter to ambiguous viral integrations
    • Additional .summary.tsv output columns
    • Added QCstatus & direct taxid counts to summary output
    • only extracting SC and OEA reads
    • EXCESSIVE_VIRAL_COVERAGE & ASSEMBLY_ABORTED QC failure modes
    • Hard filtering < 10% viral coverage
    • Actually using precomputed metrics in GRIDSS calling
  • New ubuntu 20 Dockerfile mostly based on @alexiswl #461
  • Added error message pointing to the log file with the underlying cause of the error
  • Added examples/annotate_most_likely_centromere.R
  • #439 extended documentation of GRIDSS SAM tags
  • #448 added gridsstools source code to release package so the bioconda recipe can rebuild from source
  • #450 stabilising sort order in VCF output
  • #450 setting bwa mem batch size (-K) to force deterministic behaviour
  • #449 preventing underflow of contig bounds
  • #444 fixed handling of --useproperpair
  • #463 fixed race in assembly bam header creation
  • Internally treating flanking indels are clipping (e.g. 4I5M becomes 4S5M)
    • Reads containing only clipping and I/D alignments are treated as unmapped (bwa occasionally reports alignments such as 36S30I85S)
  • Added reference matching checks to start of GRIDSS
  • Updated ComputeSamTags behaviour
    • Dropping excessively overlapping split read alignments (contained alignment & >25bp overlaps)
    • #282 Add summary stats to ComputeSamTags/PreprocessForBreakendAssembly output
    • #363 preferring non-ALT alignments when determining best split reads
    • Added summary output file containing tag changes
  • #438 added --otherjvmheap
  • Fixed bug causing assembly recovery to fail if near (<50kbp) end of chromosome