cschanhniem · Jul 9, 2026 · Jul 9, 2026
diff --git a/Makefile b/Makefile
@@ -1,4 +1,4 @@
-.PHONY: help demo test lint ci clean bench-leakage bench-multi-negatives bench-baseline bench-hidden-active bench-cluster-split bench-expert-ablation bench-expert-ablation-500 bench-selectivity bench-feature-decomp bench-gate bench-easy-baseline bench-charge-matched bench-order-dependent bench-precision-at-k bench-per-family bench-simulation-ablation bench-simulation-ablation-within-amp simulation-gate regenerate-all generate phase3 pilot validate-scoring validate-scoring-phase3 validate-scoring-strict external-pilot pilot-confident presynth-qc gold-standard diversity synthesis-order novelty-broad external-consensus questionnaire gate-check ip-report benchmark-card wave0-5-gate-check wave0-5-novelty-audit wave0-5-novelty-audit-v2 wave0-5-panel wave0-5-evidence wave0-5-fill-external wave0-5b-generate wave0-5b-filter recalibration-engine recalibration-engine-dry-run validate-policy-version generate-synthetic-lab-results bump-policy-version generate-review-packet failed-candidate-report safe-publication-filter classify-negative-informativeness
+.PHONY: help demo test lint ci clean bench-leakage bench-multi-negatives bench-baseline bench-hidden-active bench-cluster-split bench-expert-ablation bench-expert-ablation-500 bench-selectivity bench-feature-decomp bench-gate bench-easy-baseline bench-charge-matched bench-order-dependent bench-precision-at-k bench-per-family bench-simulation-ablation bench-simulation-ablation-within-amp simulation-gate regenerate-all generate phase3 pilot validate-scoring validate-scoring-phase3 validate-scoring-strict external-pilot pilot-confident presynth-qc gold-standard diversity synthesis-order novelty-broad external-consensus questionnaire gate-check ip-report benchmark-card wave0-5-gate-check wave0-5-novelty-audit wave0-5-novelty-audit-v2 wave0-5-panel wave0-5-evidence wave0-5-fill-external wave0-5b-generate wave0-5b-filter recalibration-engine recalibration-engine-dry-run validate-policy-version generate-synthetic-lab-results bump-policy-version generate-review-packet failed-candidate-report safe-publication-filter classify-negative-informativeness validate-rejection-events negative-result-dashboard
 
 PYTHON := $(shell [ -f .venv/bin/python ] && echo .venv/bin/python || echo python3)
 PYTEST  := $(shell [ -f .venv/bin/pytest ] && echo .venv/bin/pytest || echo pytest)
@@ -620,3 +620,9 @@ validate-rejection-events:
 		--input examples/rejection_events_example.json \
 		--out-json outputs/rejection_events_validation.json \
 		--out-md outputs/rejection_events_validation.md
+
+negative-result-dashboard:
+	PYTHONPATH=src $(PYTHON) scripts/negative_result_dashboard.py \
+		--input examples/negative_result_dashboard_example.json \
+		--out-json outputs/negative_result_dashboard.json \
+		--out-md outputs/negative_result_dashboard.md
diff --git a/docs/evidence/METRICS_CURRENT.md b/docs/evidence/METRICS_CURRENT.md
@@ -5,7 +5,9 @@ Machine-readable snapshot: `outputs/metrics_snapshot.json` regenerated with `mak
 > **Purpose:** One authoritative table of current pipeline metrics. If any doc disagrees
 > with this file, this file wins. Updated whenever benchmark/benchmark config changes.
 >
-> **Last updated:** 2026-07-09 (Phase F F8 — bulk rejection-event validator — v0.5.79)
+> **Last updated:** 2026-07-09 (Phase F F9 — negative-result dashboard — v0.5.79)
+> **New in v0.5.79:** Negative-result dashboard (F9) — CLI (`scripts/negative_result_dashboard.py`) that reads a collection of negative-result entries from a JSON file and produces a structured dashboard with summary statistics (by category, by pipeline version), score distributions (activity, safety, novelty, ensemble), per-category cross-analysis, and pipeline insights (most common failure category, recalibration opportunities). Outputs JSON + Markdown. Example file (`examples/negative_result_dashboard_example.json`) with 15 toy entries across all 6 failure categories. Schema (`schemas/negative_result_dashboard.schema.json`). `make negative-result-dashboard` target. 33 tests. 2883 total.
+>
 > **New in v0.5.79:** Bulk rejection-event validator (F8) — CLI (`scripts/validate_rejection_events.py`) that reads a JSON list of rejection events, validates each `rejection_code` against the rejection taxonomy (`examples/rejection_taxonomy_example.json`), checks required fields (candidate_id, rejection_code, date, pipeline_version), and outputs a PASS/FAIL validation report with per-event errors, rejection-code distribution, and Markdown summary. Example file (`examples/rejection_events_example.json`) with 6 toy events. `make validate-rejection-events` target. 26 tests. 2849 total.
 >
 > **New in v0.5.78:** Calibration link from negative-result entries to intake reports (F7) — closes learning loop by tracing each negative-result entry back to its prediction-vs-actual data in the calibration intake report. Schema addition: `intake_report_id` optional field on `negative_result_entry.schema.json`. CLI (`scripts/link_negative_result_to_intake.py`) reads negative-result archive JSON + calibration intake report, links by candidate_id, reports matched/unmatched entries, orphan intake candidates, lab summary for matched entries, and validates intake_report_id references. Produces structured JSON + Markdown link report. 25 tests. 2822 total.
@@ -68,7 +70,7 @@ Machine-readable snapshot: `outputs/metrics_snapshot.json` regenerated with `mak
 > **New in v0.5.31:** Added dipeptide-order features for sequence-order awareness. `dipeptide_order_score` achieves AUROC 0.7861 on AMP-vs-scrambled discrimination — the strongest order-dependent feature in the pipeline. Only 7/31 features survive scrambling (amphipathicity/helix-wheel + dipeptide). All composition features are purely position-independent (exactly 0.5000 AUROC on scrambled test).
 > **New in v0.5.30:** Easy baseline benchmark added — charge density alone (AUROC 0.8166) outperforms the full pipeline ensemble (0.7792) on AMP-vs-Swiss-Prot-decoy discrimination. Honest finding documented: expected because pipeline optimizes for safety, not raw discrimination.
 > **New in v0.5.29:** Expanded benchmark to 500 AMPs + 500 composition-matched decoys (n=1000). AUROC 0.7792 (CI₉₅: 0.7505–0.8065) confirms signal generalizes. Cluster-aware CI: 0.746–0.8102. Representative AUROC: 0.778. Standard benchmark (n=191) retained for backward comparison.
-> **Pipeline version:** v0.5.79
+> **Pipeline version:** v0.5.79 (F9)
 > **Branch:** main
 
 ---

diff --git a/docs/research/ROADMAP.md b/docs/research/ROADMAP.md
@@ -1,5 +1,59 @@
 # Roadmap
 
+## v0.5.79 — Loop 79: Phase F F9 — Negative-Result Dashboard ✓ (2026-07-09)
+
+CLI that reads a collection of negative-result entries from a JSON file and
+produces a structured dashboard with summary statistics, score distributions,
+per-category cross-analysis, and pipeline insights. Shows the value of
+failures by making failure patterns reviewable at a glance.
+
+Changes:
+- `scripts/negative_result_dashboard.py` (F9) — Standalone CLI that loads
+  negative-result entries (list or dict with `entries` key), builds a dashboard
+  with category breakdown, pipeline version distribution, per-score statistics
+  (mean, min, max, std), per-category score cross-analysis, and insights (most
+  common failure category, recalibration opportunities). Supports `--out-json`
+  and `--out-md` output. Exit 0 on success, 2 on input errors.
+- `schemas/negative_result_dashboard.schema.json` (F9) — JSON Schema Draft
+  2020-12 for the dashboard output. Validates dashboard_metadata, summary
+  (by_category, by_pipeline_version), scores_distribution (activity, safety,
+  novelty, ensemble with mean/min/max/std), category_score_summary, insights
+  (most_common_failure_category, recalibration_opportunities), and _caveat.
+- `examples/negative_result_dashboard_example.json` (F9) — Toy example with 15
+  entries spanning all 6 failure categories (pre_selection_reject, selected_untested,
+  lab_inactive, lab_toxic, control_failure, synthesis_failure) across 9 pipeline
+  versions. Clearly marked EXAMPLE — NOT REAL DATA.
+- `Makefile` — Added `negative-result-dashboard` target.
+- `tests/evidence/test_negative_result_dashboard.py` — 33 tests covering:
+  dashboard required structure, summary counts, by-category breakdown,
+  by-pipeline-version distribution, score distribution stats (min, max, mean,
+  count), insights (most common category, highest activity category, lowest
+  safety category, recalibration opportunities), empty entries handling,
+  entries without scores, category score summary shape, markdown sections
+  (title, summary, category table, score distribution, insights, caveat),
+  example file loading, list/dict input formats, missing file error, empty
+  list error, missing required fields error, CLI exit codes (0, 2), JSON
+  output writing, markdown output writing, and example file round-trip.
+- `docs/evidence/METRICS_CURRENT.md` — v0.5.79 F9 changelog. Test count: 2883.
+- `tests/test_test_count_regression.py` — baseline updated to 2883.
+
+Honest boundaries:
+- Dashboard summaries are computational aggregations of pipeline failure
+  records, not biological measurements.
+- A category with many entries may reflect pipeline prioritisation choices
+  (e.g., more candidates pass through pre-selection than reach lab testing)
+  rather than intrinsic biological failure rates.
+- Score distributions reflect pipeline-predicted scores, not measured
+  activity, safety, or novelty.
+- The "most common failure category" insight is a descriptive count, not
+  evidence that the pipeline is failing correctly or incorrectly.
+- Recalibration opportunity counts track entries flagged with
+  `recalibration_used=yes`; they do not measure whether recalibration
+  actually improved predictions.
+- The dashboard does not determine whether rejected candidates would have
+  been biologically active or safe. All conclusions require qualified
+  human review and wet-lab validation.
+
 ## v0.5.79 — Loop 78: Phase F F8 — Bulk Rejection-Event Validator ✓ (2026-07-09)
 
 CLI that validates rejection events against the rejection taxonomy. Checks

diff --git a/examples/negative_result_dashboard_example.json b/examples/negative_result_dashboard_example.json
@@ -0,0 +1,252 @@
+{
+  "schema": "EXAMPLE — NOT REAL DATA. For demonstration only.",
+  "description": "Toy negative-result entries for dashboard demonstration. All candidates, sequences, scores, and reasons are synthetic and must not be treated as real experimental results.",
+  "pipeline_version": "v0.5.79",
+  "entries": [
+    {
+      "entry_id": 101,
+      "date": "2026-06-01",
+      "candidate_id": "DASH_EXAMPLE-001",
+      "sequence": "RRIRIIRRIRIIRRI",
+      "reason_category": "pre_selection_reject",
+      "reason_detail": "Sequence exceeds maximum allowed charge threshold of +6 (net charge +8). Flagged by pre-selection charge filter.",
+      "pipeline_version": "v0.5.70",
+      "source_batch": "wave0.5",
+      "score_activity": 0.82,
+      "score_safety": 0.35,
+      "score_novelty": 0.55,
+      "score_ensemble": 0.72,
+      "recalibration_used": "no"
+    },
+    {
+      "entry_id": 102,
+      "date": "2026-06-03",
+      "candidate_id": "DASH_EXAMPLE-002",
+      "sequence": "GLLRRLRDFLKKLG",
+      "reason_category": "pre_selection_reject",
+      "reason_detail": "Patent near-match flagged (Tanimoto similarity 0.89 to patent WO2020XXXXX). Referred for IP review.",
+      "pipeline_version": "v0.5.71",
+      "source_batch": "wave0.5",
+      "score_activity": 0.79,
+      "score_safety": 0.72,
+      "score_novelty": 0.31,
+      "score_ensemble": 0.68,
+      "recalibration_used": "no"
+    },
+    {
+      "entry_id": 103,
+      "date": "2026-06-05",
+      "candidate_id": "DASH_EXAMPLE-003",
+      "sequence": "KLAKLAKKLAKLAK",
+      "reason_category": "selected_untested",
+      "reason_detail": "Candidate selected for Wave 0.5 panel but deprioritised due to budget constraints. Remains in pool for future waves.",
+      "pipeline_version": "v0.5.72",
+      "source_batch": "wave0.5",
+      "score_activity": 0.85,
+      "score_safety": 0.90,
+      "score_novelty": 0.48,
+      "score_ensemble": 0.79,
+      "recalibration_used": "no"
+    },
+    {
+      "entry_id": 104,
+      "date": "2026-06-10",
+      "candidate_id": "DASH_EXAMPLE-004",
+      "sequence": "RWQWRWQWRWQWRW",
+      "reason_category": "lab_inactive",
+      "reason_detail": "MIC > 128 ug/mL against E. coli, S. aureus, and P. aeruginosa. Positive control (polymyxin B) showed expected MIC 1-2 ug/mL. Three biological replicates all negative.",
+      "pipeline_version": "v0.5.73",
+      "source_batch": "wave0.5",
+      "assay_type": "MIC",
+      "assay_result": ">128",
+      "assay_unit": "ug/mL",
+      "score_activity": 0.76,
+      "score_safety": 0.88,
+      "score_novelty": 0.42,
+      "score_ensemble": 0.71,
+      "recalibration_used": "no"
+    },
+    {
+      "entry_id": 105,
+      "date": "2026-06-12",
+      "candidate_id": "DASH_EXAMPLE-005",
+      "sequence": "GIGKFLHSAKKWGKAFVGEIMNS",
+      "reason_category": "lab_inactive",
+      "reason_detail": "MIC > 256 ug/mL against all tested strains. Magainin-2 analog with single-point mutations expected to retain activity but showed complete loss. Structural analysis pending.",
+      "pipeline_version": "v0.5.73",
+      "source_batch": "wave0.5",
+      "assay_type": "MIC",
+      "assay_result": ">256",
+      "assay_unit": "ug/mL",
+      "score_activity": 0.81,
+      "score_safety": 0.85,
+      "score_novelty": 0.39,
+      "score_ensemble": 0.74,
+      "recalibration_used": "no"
+    },
+    {
+      "entry_id": 106,
+      "date": "2026-06-15",
+      "candidate_id": "DASH_EXAMPLE-006",
+      "sequence": "FLPIVAGKLLKGL",
+      "reason_category": "lab_toxic",
+      "reason_detail": "HC50 = 8 ug/mL (hemolytic). Hemolysis > 50% at 25 ug/mL. Flagged as unsafe for systemic administration.",
+      "pipeline_version": "v0.5.74",
+      "source_batch": "wave0.5",
+      "assay_type": "hemolysis_RBC",
+      "assay_result": "8",
+      "assay_unit": "ug/mL HC50",
+      "score_activity": 0.88,
+      "score_safety": 0.42,
+      "score_novelty": 0.52,
+      "score_ensemble": 0.76,
+      "recalibration_used": "yes"
+    },
+    {
+      "entry_id": 107,
+      "date": "2026-06-18",
+      "candidate_id": "DASH_EXAMPLE-007",
+      "sequence": "KWRRWRRWKWRRWR",
+      "reason_category": "lab_toxic",
+      "reason_detail": "Cytotoxicity CC50 = 4 ug/mL against HEK293 cells. Selectivity index < 1.0. High membrane disruption activity confirmed.",
+      "pipeline_version": "v0.5.74",
+      "source_batch": "wave0.5",
+      "assay_type": "cytotoxicity_mammalian",
+      "assay_result": "4",
+      "assay_unit": "ug/mL CC50",
+      "score_activity": 0.91,
+      "score_safety": 0.28,
+      "score_novelty": 0.58,
+      "score_ensemble": 0.78,
+      "recalibration_used": "yes"
+    },
+    {
+      "entry_id": 108,
+      "date": "2026-06-20",
+      "candidate_id": "DASH_EXAMPLE-008",
+      "sequence": "AAAAKAAAAKAAAA",
+      "reason_category": "control_failure",
+      "reason_detail": "Positive control (polymyxin B) MIC = 32 ug/mL against E. coli (expected 1-2 ug/mL). Assay batch invalidated. Candidate not tested.",
+      "pipeline_version": "v0.5.74",
+      "source_batch": "wave0.5",
+      "assay_type": "MIC",
+      "score_activity": 0.45,
+      "score_safety": 0.95,
+      "score_novelty": 0.22,
+      "score_ensemble": 0.50,
+      "recalibration_used": "no"
+    },
+    {
+      "entry_id": 109,
+      "date": "2026-06-22",
+      "candidate_id": "DASH_EXAMPLE-009",
+      "sequence": "RRWQWRMKKLG",
+      "reason_category": "synthesis_failure",
+      "reason_detail": "SPPS purification yield < 5%. Multiple hydrophobic stretches caused aggregation during synthesis. Three synthesis attempts all failed.",
+      "pipeline_version": "v0.5.75",
+      "source_batch": "wave0.5",
+      "score_activity": 0.73,
+      "score_safety": 0.80,
+      "score_novelty": 0.61,
+      "score_ensemble": 0.70,
+      "recalibration_used": "no"
+    },
+    {
+      "entry_id": 110,
+      "date": "2026-06-25",
+      "candidate_id": "DASH_EXAMPLE-010",
+      "sequence": "ACDEFGHIKLMNPQRSTVWY",
+      "reason_category": "synthesis_failure",
+      "reason_detail": "Trp oxidation detected during cleavage. Crude purity < 60%. Not suitable for biological testing without significant re-optimization.",
+      "pipeline_version": "v0.5.75",
+      "source_batch": "wave0.5",
+      "score_activity": 0.68,
+      "score_safety": 0.75,
+      "score_novelty": 0.45,
+      "score_ensemble": 0.65,
+      "recalibration_used": "no"
+    },
+    {
+      "entry_id": 111,
+      "date": "2026-06-28",
+      "candidate_id": "DASH_EXAMPLE-011",
+      "sequence": "KLLKLLLKLLKLLL",
+      "reason_category": "selected_untested",
+      "reason_detail": "Candidate selected for Wave 0.5b but deferred pending safety assay optimisation. Expected to be tested in Wave 0.5c.",
+      "pipeline_version": "v0.5.76",
+      "source_batch": "wave0.5b",
+      "score_activity": 0.83,
+      "score_safety": 0.87,
+      "score_novelty": 0.35,
+      "score_ensemble": 0.75,
+      "recalibration_used": "no"
+    },
+    {
+      "entry_id": 112,
+      "date": "2026-07-01",
+      "candidate_id": "DASH_EXAMPLE-012",
+      "sequence": "RRWRRWRRWRRW",
+      "reason_category": "lab_inactive",
+      "reason_detail": "No bactericidal activity at 64 ug/mL against any of 5 tested strains. Time-kill assay confirmed static effect only at highest concentration.",
+      "pipeline_version": "v0.5.77",
+      "source_batch": "wave0.5b",
+      "assay_type": "MIC",
+      "assay_result": ">64",
+      "assay_unit": "ug/mL",
+      "score_activity": 0.80,
+      "score_safety": 0.52,
+      "score_novelty": 0.60,
+      "score_ensemble": 0.73,
+      "recalibration_used": "no"
+    },
+    {
+      "entry_id": 113,
+      "date": "2026-07-03",
+      "candidate_id": "DASH_EXAMPLE-013",
+      "sequence": "GIGAVILKVATGLPTLISWIK",
+      "reason_category": "pre_selection_reject",
+      "reason_detail": "Known toxic sequence pattern detected (melittin-like). Flagged by safety pre-filter.",
+      "pipeline_version": "v0.5.77",
+      "source_batch": "wave0.5b",
+      "score_activity": 0.87,
+      "score_safety": 0.15,
+      "score_novelty": 0.33,
+      "score_ensemble": 0.67,
+      "recalibration_used": "no"
+    },
+    {
+      "entry_id": 114,
+      "date": "2026-07-05",
+      "candidate_id": "DASH_EXAMPLE-014",
+      "sequence": "LLRLLRRLLRLLRR",
+      "reason_category": "control_failure",
+      "reason_detail": "Hemolysis control (Triton X-100) outside expected range. Assay batch flagged for QC review. Results not valid.",
+      "pipeline_version": "v0.5.78",
+      "source_batch": "wave0.5b",
+      "assay_type": "hemolysis_RBC",
+      "score_activity": 0.55,
+      "score_safety": 0.70,
+      "score_novelty": 0.30,
+      "score_ensemble": 0.52,
+      "recalibration_used": "no"
+    },
+    {
+      "entry_id": 115,
+      "date": "2026-07-07",
+      "candidate_id": "DASH_EXAMPLE-015",
+      "sequence": "RWKWRWKWRWKW",
+      "reason_category": "lab_inactive",
+      "reason_detail": "Biofilm inhibition < 25% at 128 ug/mL against S. aureus biofilm. Anti-biofilm assay showed no significant activity compared to untreated control.",
+      "pipeline_version": "v0.5.78",
+      "source_batch": "wave0.5b",
+      "assay_type": "biofilm_inhibition",
+      "assay_result": "<25",
+      "assay_unit": "% inhibition at 128 ug/mL",
+      "score_activity": 0.77,
+      "score_safety": 0.65,
+      "score_novelty": 0.55,
+      "score_ensemble": 0.71,
+      "recalibration_used": "no"
+    }
+  ]
+}