Skip to content

Releases: pieetie/svforge

v1.0.2

Choose a tag to compare

@pieetie pieetie released this 25 May 16:53
920805c

v1.0.2 - 2026-05-25

Fixed

  • DELLY writer: FORMAT field order corrected from …FT:RC:RCL:RCR… to …FT:RCL:RC:RCR… to match real DELLY 1.x output; sample column values reordered accordingly so RC stays the SV-region count and RCL/RCR the left/right control regions.
  • Manta writer: DUP records now emit <DUP:TANDEM> as ALT (matching the ##ALT declaration in the Manta header) instead of the undeclared <DUP>.
  • DELLY writer: BND records now use the standard VCF 4.2 breakend notation (e.g. ]chr3:151368767]C) instead of the symbolic <BND> allele, matching real DELLY output.
  • Sampler: SV anchor REF base is now drawn uniformly from A/C/G/T per record instead of being hardcoded to N. This makes BND ALT strings look like G]chr1:42] instead of N]chr1:42].

Added

  • README "Scope & limitations" section documenting which INFO/FORMAT fields are declared in the bundled headers but not (yet) populated, split by caller (Manta vs DELLY).

v1.0.1

Choose a tag to compare

@pieetie pieetie released this 27 Apr 23:14
f47ddf3

v1.0.1 - 2026-04-28

Fixed

  • gen-pair now produces a two-sample somatic VCF (#7).
  • VCF output is now tabix-indexable; previously inter-chromosomal BND records broke sort order (#6).
  • README DOI and license badges (#8, #9).

Changed

  • BREAKING: gen-pair replaces --out-tumor / --out-normal with a single --out flag.

v1.0.0 — Initial release

Choose a tag to compare

@pieetie pieetie released this 25 Apr 12:25
9a9bf7b

[1.0.0] — 2026-04-25

Initial release.

  • svforge gen and svforge gen-pair to generate synthetic VCFs (single-sample or coherent tumor/normal pair)
  • Manta (VCFv4.1) and DELLY (VCFv4.2) writers
  • Injection via --gnomad-fraction and --blacklist-fraction
  • INFO/SVFORGE_SOURCE tag on every record (bank / gnomad / blacklist) for self-verifying pipelines
  • svforge validate to confirm injected records match the bundled catalogs exactly
  • Chromosome filtering via --chromosomes
  • Custom header support via --header-template PATH
  • Deterministic sampling via --seed; effective seed always logged in the output VCF
  • BCF / VCF.gz / VCF output formats
  • Default hg38 SV bank with weighted templates across DEL, DUP, INV, INS, BND
  • Writer plugin system: third-party callers can register via svforge.writers entry point
  • Non-configurable ##svforgeWarning=SYNTHETIC_DATA_DO_NOT_USE_FOR_CLINICAL_DIAGNOSIS injected in every VCF header
  • sanitize_command() strips absolute paths from the logged command line so user home directories and cluster paths never leak into generated VCFs
  • Python 3.10+, hg38 only