GitHub
09/21/2026, 8:19 AM4, 19). A pipeline that renames UCSC names to Ensembl ones therefore has that step exercised only as a no-op, and a rename bug passes CI unnoticed. Real callers emit UCSC names: Mutect2 and HaplotypeCaller both write chr1, chr4, chrM.
This came out of nf-core/epitopeprediction#362, where germline context was silently dropped for every real user because the germline VCF kept its chr prefix while the somatic calls had already been renamed. The existing fixtures could not catch it; a real exome pair did.
Files
• testdata/variants/variants_chr4_chr19_ucsc.vcf — variants_chr4_chr19.vcf with chr-prefixed contigs.
• testdata/variants/variants_chr4_chr19_germline_ucsc.vcf — variants_chr4_chr19_germline.vcf with chr-prefixed contigs.
• testdata/sample_sheets/sample_sheet_variants_grch38_ucsc.csv — pairs the two.
Only the contig names change, in both the ##contig headers and the records. Same four somatic and two germline records at the same positions, everything else byte-identical.
Why this shape
The reference and VEP cache on this branch stay Ensembl-named, which is the real-world arrangement: callers emit UCSC names, Ensembl resources do not, and the pipeline bridges them. Because only the naming differs, a correctly renaming pipeline produces exactly the results the Ensembl-named fixtures produce, so a test can assert the same expected peptides and any difference is a rename failure.
🤖 Generated with Claude Code
nf-core/test-datasetsGitHub
09/21/2026, 8:22 AMraw.githubusercontent.com using a mutable ref (epitopeprediction).…> · New
• Medium severity <#discussion_r4060358813|This fixture embeds extensive absolute filesystem paths and internal project/sample identifiers…> · New
• Medium severity <#discussion_r4060358841|This fixture embeds extensive absolute filesystem paths and internal project/sample identifiers…> · New
• Medium severity <#discussion_r4060358860|This fixture embeds extensive absolute filesystem paths and internal project/sample identifiers…> · New
• Low severity <#discussion_r4060358886|The embedded provenance commands are misleading for this UCSC-named fixture: one references…> · New
• Low severity <#discussion_r4060358915|The embedded provenance commands are misleading for this UCSC-named fixture: one references…> · New
What changed in this PR
Adds UCSC (chr-prefixed) equivalents of the existing chr4/chr19 somatic + germline VCF fixtures so CI can exercise contig renaming logic against real-world caller output.
Changes:
• Added UCSC-named somatic VCF fixture for chr4/chr19 variants.
• Added UCSC-named germline VCF fixture for chr4/chr19 variants.
• Added a sample sheet entry wiring the UCSC somatic + germline fixtures together.
| File | Description |
| -------------------------------------------------------------- | --------------------------------------------------------- |
| testdata/variants/variants_chr4_chr19_ucsc.vcf | New somatic fixture with chr4/chr19 contigs and records. |
| testdata/variants/variants_chr4_chr19_germline_ucsc.vcf | New germline fixture with chr4/chr19 contigs and records. |
| testdata/sample_sheets/sample_sheet_variants_grch38_ucsc.csv | New sample sheet pointing at the UCSC fixtures. |
---
💡 </nf-core/test-datasets/new/epitopeprediction?filename=.github/skills/code-review/SKILL.md|Add a code-review agent skill> or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
nf-core/test-datasetsGitHub
09/21/2026, 8:23 AM