Skip to content

CHANGELOG.md

Latest commit

 

History

History
122 lines (98 loc) · 7.09 KB

CHANGELOG.md

File metadata and controls

122 lines (98 loc) · 7.09 KB

v0.10.5

Fixed

  • Updated dependencies to remove known vulnerabilities

v0.10.4

Changes

  • Exposes a new option --max-branch-factor in both compare and merge modes. This option controls a heuristic that limits the exploration of the sequence search space during query optimization. Increasing the value may lead to slightly more accurate results at the cost of increased compute requirements.

v0.10.3

Fixed

  • Improved error reporting from VCF parsing to indicate the most recently parsed record
  • Fixed an off-by-one error in the output of the merge region BED files

v0.10.2

Changes

  • Improved inline documentation that appears on docs.rs

Fixed

  • Updates the GitHub workflows to patch an issue with tarball version naming

v0.10.1

Fixed

  • Fixed a versioning issue with dependencies

v0.10.0

Changes

  • GitHub release workflow updated to auto-publish to crates.io
  • The library has been renamed to aardvark-bio for publishing to crates.io, binary name is unchanged

v0.9.0

Changes

  • Adds the optional RECORD_BP comparison type to all summary files. Review our methods for details on this metric. This metric is enabled with --enable-record-basepair-metrics.
  • The GT and BASEPAIR metrics are recommend for most users. Thus, the HAP and WEIGHTED_HAP metrics are now disabled by default. Two new options have been added to enable these secondary metrics: --enable-hapotype-metrics and --enabled-weighted-haplotype-metrics.
  • Refactored the way group metrics are internally represented for easier long-term maintenance
  • Added new GitHub automations for developer ease-of-use

v0.8.1

Changes

  • Adds automated workflow for static binary build and release. Aardvark functionality identical to v0.8.0.

v0.8.0

Changes

  • Adds beta support for medium-sized variant types (<= 10 kbp):
    • SvDeletion and SvInsertion - Structural variants identified by the SVTYPE tag in the provided VCF files
    • TrContraction and TrExpansion - Tandem repeat variants identified by a TRID tag in the provided VCF files
    • Add JointStructuralVariant and JointTandemRepeat types, which are similar to JointIndel
    • Including these larger variant types increase the compute cost of Aardvark
  • Adds additional heuristic limitation to reduce compute costs from larger variant types:
    • Adds a filter on variants larger than 10 kbp
    • Adds a short-circuit for GT candidate scoring such that any edits immediately ends the candidate exploration
    • Adds a limit on the number of candidates that can get generated before a sync point is identified in GT scoring, preventing exponentially blow-up of GT scoring when large variants mismatch
    • Adds filter on variants that are not sequence resolved (e.g., "" entry is not supported)

Fixed

  • Updates noodles crate to resolve some parsing issues:
    • Fixed issue with parsing SVLEN field
    • Fixed a VCF parsing issue where absent chromosomes would error instead of gracefully returning no variants
  • Updated build to rust:1.88.0 to resolve some compilation issues

v0.7.4

Changes

  • Adds a new scoring mode, WEIGHTED_HAP. This scoring mode is similar to HAP scoring, but variants are weighted by the number of changes between the REF and ALT sequences. For SNPs, the HAP and WEIGHTED_HAP scores should be identical since all SNPs have the same weight. For indels, each variant is effectively weighted by its length, so longer variants have an increased weight.

Fixed

  • Removed an exact-match shortcut in compare mode that would occasionally under-estimate the variant-level errors

v0.7.3

Changes

  • Added REF/ALT trimming of identical tail basepair sequences, which resolves some edge-case variant conflicts in the T2T truth set. This minor change tends to slightly improve the overall accuracy by reducing variant conflicts, allowing for overlapping changes when the trimming removes previously conflicting bases. Some Indel variants may be classified differently compared to previous versions. See the output VCF files for trimmed representations. Examples:
    • AC->CC : this is trimmed to A->C and classified as a SNV now.
    • ACC->AC : this is trimmed to AC->A and classified as a Deletion now.
    • AC->ACC : this is trimmed to A->AC and classified as an Insertion now.
  • Adds a new option --disable-variant-trimming to both compare and merge modes, which disable the above trimming behavior.
  • Updated the install documentation to reflect bioconda support

v0.7.2

Fixed

  • Added changes to build script to enable bioconda building from source

v0.7.1

Changes

  • Added two new GT-specific statistics to the summary files:
    • truth_fn_gt - The number of truth_fn where the allelic sequence was matched in both inputs, but with the wrong genotype (e.g., 1/1 in truth, 0/1 in query). This value is only populated if the comparison is GT.
    • query_fp_gt - The number of query_fp where the allelic sequence was matched in both inputs, but with the wrong genotype (e.g., 0/1 in truth, 1/1 in query). This value is only populated if the comparison is GT.
  • Added parallelization to the writers for both compare and merge.

v0.7.0

Changes

  • Added a new option to compare mode: --stratifications. If provided, this will post-annotate all regions by the provided stratifications and add additional rows to the output summary TSV, see documentation for more details.
  • Added query_total to the main summary and region summary output files. This metric is query_tp + query_fp, and is analogous to truth_total.
  • Replaced the info statements during file writing with a progress bar.

Fixed

  • Fixed an issue where "*" ALT alleles were treated as alternate sequence. They are now ignored.

v0.6.0

Changes

  • In compare mode, the --confidence-regions parameter has been replaced with --regions for consistency.
  • The progress bar has been added to the parallelized variant loading step.

Fixed

  • Fixed an issue where variants that were hemizygous would be ignored entirely. They are now treated as homozygous variants for comparison.

v0.5.4

Changes

  • Added a short-circuit in the merge routine that checks variant lengths prior to trying a full comparison. This reduces run-time significantly when larger events are unique to an input.
  • Adds parallelization to the file loading for both compare and merge. The parallelization is across both file and chromosome for the typical process, significantly reducing initially variant loading times.

v0.5.3

Changes

  • Added a new optional output file for merging (--output-summary) which contains statistics on the merge. See documentation for details.

v0.5.2

Changes

  • Added a new file (region_sequences.tsv.gz) to the debug output for aardvark compare. This file contains the constructed haplotype sequences for each region.
  • Replaced the region summary file in the debug output with a gzip-compressed version (region_summary.tsv -> region_summary.tsv.gz) for aardvark compare. In internal tests, this reduced the disk footprint by ~90% for this file.

Fixed

  • Updated crossbeam-channel for security update

v0.5.1

Initial release.