Pfam proposals, CGD curation leads, owner decisions, per-call status spec (grilling session 2026-10-07)

[hyphaltip]

What this adds

Output of the owner grilling session of 2026-10-07. Handoff: docs/HANDOFF-2026-10-07.md (decisions 1 to 12 are in section 2).

• Pfam proposals. 122 hits in 6 proteomes, each marked proposed (docs/reports/data/sorting_hat/specificity/pfam_proposals.tsv, analysis/sorting_hat_run/pfam_proposals.py). Family-level proposal: docs/reports/2026-10-07-pfam-family-proposals.md. No family is set active. data/sorting_hat/family_table.tsv is unchanged.
• Owner decision on Q6FTA2 and Q59TP1 (RBT1): excluded from the repeat controls only (owner_decisions.tsv, adjudication README). RBT1 keeps its adhesin E2 label with PMIDs 10978273, 19837954, 21414038. Braun 2000 was read: the two RBT1 alleles differ by a segment (aa 612 to 640), not by repeat number.
• R0 status source added to the two S288C runs (both now smoke). Report and run copies updated.
• FLO11 repeat miss explained (coverage 0.045 against the 0.25 cut). The FLO11 statement in analysis/cocci_repeats/REPORT_2026-09-30_pfam_confirmation.md is not supported by this measurement; that file is not changed.
• Curation leads from CGD: docs/curation-leads/, data/curated/biofilm/cgd_biofilm_literature_leads.tsv, cgd_adherence_literature_leads.tsv.
• Spec for per-call status entries (docs/superpowers/specs/2026-10-07-per-call-status-design.md). Draft. No independent review yet and no code.
• Verification: YPS3 accession (name matches), BcLysM1 sentence (supported by the abstract), WHO/IUIS counts (recount of the stored table; allergen.org not queried), table B3 of docs/paper/02 (24 values match metrics.json; the B1 cell for A. nidulans is filled).

Read before merging

• The Pfam classes are proposals from table facts, not expert reviews. 48 of 122 are needs_expert.
• Pfam sign-off still needs: SignalP on C. neoformans H99, a C. albicans run, a decision on AltA1, an expert pass, and reviewed member lists (issue Step 2b/2c: reviewed member and non-member controls and specificity tests for the 15 Pfam families (task 04) #69). None of these runs was done here.
• Not verified: the three cell wall reviews, the other "copied" rows of docs/paper/02.
• The STAR job of 2026-10-03 failed (BAM sorting memory); the handoff says so. It was not rerun.

Pre-commit (ruff 0.3.5 and ruff-format) passes. pfam_proposals.py reproduces the stored TSV exactly. No test was added: the scripts are analysis scripts.

🤖 Generated with Claude Code