Point INTERVIEW→GALAXY scenarios at gap-closing reference workflows

[jmchilton]

Posted by Claude (AI assistant) on behalf of jmchilton.

Both INTERVIEW → GALAXY scenario targets were deliberately known-incomplete extractions. This swaps each for the iterated workflow that closes its catalogued gap, and loosens the WIP hedging accordingly.

What changed

UC1 / MRSA — UC1_MRSA_extracted.ga (14-step) → UC1_MRSA_Bakta_JBrowse_faithful.ga (35-step).

• The docs name this "the definitive artifact … re-runs byte-identical to golden." clean_extract is explicitly superseded (structurally clean but map→reduce unfaithful); extracted is the earliest.
• Integrates Bakta whole-genome annotation + per-isolate JBrowse locus view — the two deliverables the old target dropped. Six __BUILD_LIST__ nests keep every mapped multiple-input (staramr, both closest joins, three JBrowse tracks) from silently reducing.
• Note: this diverges from the paper SI, which kept the clean 14-step S1 as its published baseline. Chosen as a richer compare reference.

UC3 / ATAC — UC3_ATAC_extracted.ga (5-step) → UC3_ATAC_extracted_figures.ga (13-step, paper SI Workflow S5).

• Materializes the significance filter → Sort → Select first/Select last (top-gained / top-lost) ranked tables as real workflow outputs, instead of trapping the differential call inside the volcano PDF's coloring.

Hedging loosened in scenarios.md + issue headers

• Dropped "not human-validated", "known-incomplete by design / they aren't good enough yet", and both per-case "known limitation (bonus points to beat it)" blocks.
• Kept the deliberate "compare-and-contrast reference, not a hard oracle / divergence is the signal" stance — that's design intent, not WIP hedging.
• Removed the two now-obsolete UC*_NEXT_STEPS.md gap-handoff notes.

Validation

npm run validate → 0 errors (172 pre-existing warnings, none in this pipeline).

🤖 Generated with Claude Code