Gene fusions: classification equivalence/discordance and breakend (BND) representation, deferred from #1506
- Dominant language
- Python
- Stars
- 30
- Forks
- 3
- Avg merge
- 9h 28m
- Merged PRs (30d)
- 42
Description
🤖 Written by Claude.
The parts of #1506 deliberately deferred when its ingestion half landed (PR #1719 and the #1558 grid work), recorded here so the TSO 500 plan (SACGF/variantgrid_sapath#431) can close. Neither is scheduled; both are "when a consumer appears".
## Fusion classification equivalence and discordance
The research-level part of #1506 that the user group's feedback was about. Fusion identity is the gene pair (`GeneFusion` on a `Variant` anchored on the gene-level contig, `snpdb/gene_level_variants.py`), with breakpoints per observation in `CohortGenotype.info` — so two labs classifying `BCR::ABL1` already land on the same variant, but nothing yet says when two fusion classifications agree or disagree. Somatic classifications get `MULTIPLE_RECORDS_DISCORDANCE_NOT_SUPPORTED` today, so nothing regresses by waiting. Nomenclature is VICC's `GENE1::GENE2`; the single-hyphen form means a read-through transcript.
## Breakend representation and BND VCF export (#1506 phases 2 and 3)
Every AllFusions.csv row carries both breakpoints (`chr8:128806980` form) and Phase 5 stores them, so this is a read-side change: representing the stored breakpoints as breakends and exporting them as BND records. Whether VEP parses BND ALT syntax at all is unverified — a 20-minute experiment before anyone designs around either answer — but what VEP would give is per-position feature overlap rather than frame or domain analysis, and the caller already reports that as `Gene A/B Location`.
Contributor guide
No contributing guide indexed for this repository
Research direction
Start with snpdb/gene_level_variants.py and the GeneFusion/CohortGenotype.info description, then review the deferred phases in #1506 and the existing AllFusions.csv and Phase 5 breakpoint handling. First verify whether VEP parses BND ALT syntax; completion requires defined fusion-classification equivalence/discordance behavior and a decided representation and export path for stored breakpoints.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- bioinformatics
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Active
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100