feat: enforce config filters, add run manifest, bench leakage CLI, expand tests

- Enforce min_length/max_length from config in score_candidates()
- Apply min_novelty and max_safety_risk selection thresholds from config
- Add `valid` field to ScoredCandidate; mark invalid sequences with failure reasons
- Add `selected` boolean to JSONL output rows
- Generate run_manifest.json with run_id, input SHA-256 hashes, config hash, pipeline version
- Add `openamp-foundry bench leakage` CLI subcommand and `make bench-leakage` target
- Improve report disclaimer to explicitly state no antimicrobial activity demonstrated
- Expand test suite from 6 to 37 tests covering: pipeline filters, selection thresholds,
  run manifest generation, benchmark leakage detection, splits, evaluation, CLI integration

Author: cschanhniem

Makefile

@@ -1,13 +1,29 @@
-.PHONY: demo test lint clean
+.PHONY: demo test lint clean bench-leakage
+
+PYTHON := $(shell [ -f .venv/bin/python ] && echo .venv/bin/python || echo python3)
+PYTEST  := $(shell [ -f .venv/bin/pytest ] && echo .venv/bin/pytest || echo pytest)
+RUFF    := $(shell [ -f .venv/bin/ruff ] && echo .venv/bin/ruff || echo ruff)
 
 demo:
-	PYTHONPATH=src python -m openamp_foundry.cli rank --candidates examples/sequences/demo_candidates.csv --references examples/known_reference/demo_known_amps.csv --out outputs/demo_ranked.jsonl --report outputs/demo_report.md --cert-dir outputs/evidence
+	PYTHONPATH=src $(PYTHON) -m openamp_foundry.cli rank \
+		--candidates examples/sequences/demo_candidates.csv \
+		--references examples/known_reference/demo_known_amps.csv \
+		--out outputs/demo_ranked.jsonl \
+		--report outputs/demo_report.md \
+		--cert-dir outputs/evidence \
+		--manifest outputs/run_manifest.json
 
 test:
-	pytest -q
+	$(PYTEST) -q
 
 lint:
-	ruff check src tests scripts
+	$(RUFF) check src tests scripts
+
+bench-leakage:
+	PYTHONPATH=src $(PYTHON) -m openamp_foundry.cli bench leakage \
+		--candidates examples/sequences/demo_candidates.csv \
+		--references examples/known_reference/demo_known_amps.csv \
+		--out outputs/leakage_report.json
 
 clean:
-	rm -rf outputs/*.jsonl outputs/*.md outputs/evidence .pytest_cache .ruff_cache
+	rm -rf outputs/*.jsonl outputs/*.md outputs/*.json outputs/evidence .pytest_cache .ruff_cache

src/openamp_foundry/cli.py

@@ -19,12 +19,22 @@ def build_parser() -> argparse.ArgumentParser:
     rank.add_argument("--out", required=True)
     rank.add_argument("--report", required=False)
     rank.add_argument("--cert-dir", required=False)
+    rank.add_argument("--manifest", required=False)
     rank.add_argument("--config", default="configs/pipeline.yaml")
 
     validate = sub.add_parser("validate", help="Validate a candidate certificate against JSON schema.")
     validate.add_argument("--certificate", required=True)
     validate.add_argument("--schema", required=True)
 
+    bench = sub.add_parser("bench", help="Run benchmark and leakage checks.")
+    bench_sub = bench.add_subparsers(dest="bench_command", required=True)
+
+    leakage = bench_sub.add_parser("leakage", help="Find near-duplicate candidates in references.")
+    leakage.add_argument("--candidates", required=True)
+    leakage.add_argument("--references", required=True)
+    leakage.add_argument("--threshold", type=float, default=0.90)
+    leakage.add_argument("--out", required=False, help="Optional JSON output path.")
+
     return parser
 
 
@@ -40,6 +50,7 @@ def main(argv: list[str] | None = None) -> int:
             report_path=args.report,
             cert_dir=args.cert_dir,
             config_path=args.config,
+            manifest_path=args.manifest,
         )
         print(json.dumps({"status": "ok", "out": args.out, "report": args.report}, indent=2))
         return 0
@@ -50,9 +61,39 @@ def main(argv: list[str] | None = None) -> int:
         print(json.dumps({"status": "valid", "certificate": args.certificate}, indent=2))
         return 0
 
+    if args.command == "bench":
+        return _run_bench(args)
+
     parser.error("unknown command")
     return 2
 
 
+def _run_bench(args: argparse.Namespace) -> int:
+    from openamp_foundry.benchmark.leakage import find_near_duplicates
+    from openamp_foundry.data.loaders import load_candidates_csv
+    from openamp_foundry.utils.io import write_json
+
+    if args.bench_command == "leakage":
+        candidates = load_candidates_csv(args.candidates)
+        references = load_candidates_csv(args.references)
+        hits = find_near_duplicates(candidates, references, threshold=args.threshold)
+        result = {
+            "status": "ok",
+            "threshold": args.threshold,
+            "near_duplicate_count": len(hits),
+            "near_duplicates": hits,
+            "warning": (
+                "Near-duplicates detected. If these candidates were used for training or "
+                "scoring baseline models, benchmark results may be inflated."
+            ) if hits else None,
+        }
+        if args.out:
+            write_json(args.out, result)
+        print(json.dumps(result, indent=2))
+        return 0
+
+    return 2
+
+
 if __name__ == "__main__":
     raise SystemExit(main())

src/openamp_foundry/pipeline.py

@@ -1,8 +1,12 @@
 from __future__ import annotations
 
+import hashlib
+import uuid
+from datetime import datetime, timezone
 from pathlib import Path
 from typing import Any
 
+from openamp_foundry import __version__
 from openamp_foundry.config import load_config
 from openamp_foundry.data.loaders import is_valid_sequence, load_candidates_csv
 from openamp_foundry.evidence.certificate import build_certificate
@@ -15,23 +19,34 @@
 from openamp_foundry.selection.diversity import greedy_diverse_select
 from openamp_foundry.selection.pareto import rank_candidates
 from openamp_foundry.types import ScoredCandidate
+from openamp_foundry.utils.hashing import file_sha256, stable_json_hash
 from openamp_foundry.utils.io import write_json, write_jsonl
 
 
+def _passes_length_filter(sequence: str, min_length: int, max_length: int) -> bool:
+    return min_length <= len(sequence) <= max_length
+
+
 def score_candidates(
     candidate_path: str | Path,
     reference_path: str | Path | None = None,
     config_path: str | Path = "configs/pipeline.yaml",
 ) -> tuple[list[ScoredCandidate], dict[str, Any]]:
     config = load_config(config_path)
-    allowed = set(config["filters"]["allowed_amino_acids"])
+    filters = config.get("filters", {})
+    allowed = set(filters.get("allowed_amino_acids", "ACDEFGHIKLMNPQRSTVWY"))
+    min_length = int(filters.get("min_length", 8))
+    max_length = int(filters.get("max_length", 35))
+
     candidates = load_candidates_csv(candidate_path)
     references = load_candidates_csv(reference_path) if reference_path else []
     weights = config["weights"]
 
     scored: list[ScoredCandidate] = []
     for candidate in candidates:
-        valid = is_valid_sequence(candidate.sequence, allowed=allowed)
+        valid_aa = is_valid_sequence(candidate.sequence, allowed=allowed)
+        valid_len = _passes_length_filter(candidate.sequence, min_length, max_length)
+        valid = valid_aa and valid_len
         features = compute_features(candidate.sequence)
         act = activity_likeness_score(features) if valid else 0.0
         safe = safety_score(features) if valid else 0.0
@@ -51,32 +66,82 @@ def score_candidates(
             references_checked=[str(reference_path)] if reference_path else [],
             nearest_reference=nearest,
         )
+        item.valid = valid
         item.selection_reason = selection_reasons(raw_scores)
         item.known_failure_modes = known_failure_modes(raw_scores)
+        if not valid_aa:
+            item.known_failure_modes.append("Sequence contains non-canonical amino acids.")
+        if not valid_len:
+            item.known_failure_modes.append(
+                f"Sequence length {len(candidate.sequence)} outside filter range "
+                f"[{min_length}, {max_length}]."
+            )
         scored.append(item)
     return scored, config
 
 
+def build_run_manifest(
+    run_id: str,
+    config: dict[str, Any],
+    input_paths: list[Path],
+    output_paths: list[str],
+    generated_at: str,
+) -> dict[str, Any]:
+    inputs = []
+    for p in input_paths:
+        if p.exists():
+            inputs.append({"path": str(p), "sha256": file_sha256(p)})
+        else:
+            inputs.append({"path": str(p)})
+    return {
+        "run_id": run_id,
+        "pipeline_version": __version__,
+        "config_hash": stable_json_hash(config),
+        "generated_at": generated_at,
+        "inputs": [x["path"] for x in inputs],
+        "input_hashes": {x["path"]: x.get("sha256", "N/A") for x in inputs},
+        "outputs": output_paths,
+    }
+
+
 def run_ranking_pipeline(
     candidate_path: str | Path,
     reference_path: str | Path | None,
     out_path: str | Path,
     report_path: str | Path | None = None,
     cert_dir: str | Path | None = None,
     config_path: str | Path = "configs/pipeline.yaml",
+    manifest_path: str | Path | None = None,
 ) -> list[ScoredCandidate]:
+    run_id = str(uuid.uuid4())
+    generated_at = datetime.now(timezone.utc).isoformat()
+
     scored, config = score_candidates(candidate_path, reference_path, config_path)
+    selection_cfg = config.get("selection", {})
+    min_novelty = float(selection_cfg.get("min_novelty", 0.0))
+    max_safety_risk = float(selection_cfg.get("max_safety_risk", 1.0))
+
     ranked = rank_candidates(scored)
-    top_n = int(config.get("selection", {}).get("top_n", len(ranked)))
-    selected = greedy_diverse_select(ranked, top_n=top_n)
+    top_n = int(selection_cfg.get("top_n", len(ranked)))
+
+    eligible = [
+        item for item in ranked
+        if getattr(item, "valid", True)
+        and item.scores["novelty"] >= min_novelty
+        and item.scores["safety"] >= (1.0 - max_safety_risk)
+    ]
+    selected = greedy_diverse_select(eligible, top_n=top_n)
 
     rows = []
+    selected_ids = {item.candidate.candidate_id for item in selected}
     for item in ranked:
         rows.append(
             {
                 "candidate_id": item.candidate.candidate_id,
                 "sequence": item.candidate.sequence,
                 "source": item.candidate.source,
+                "valid": getattr(item, "valid", True),
+                "selected": item.candidate.candidate_id in selected_ids,
                 "scores": item.scores,
                 "features": item.features,
                 "nearest_reference": item.nearest_reference,
@@ -95,6 +160,28 @@ def run_ranking_pipeline(
     if report_path:
         write_report(report_path, ranked, selected)
 
+    output_paths = [str(out_path)]
+    if report_path:
+        output_paths.append(str(report_path))
+    if cert_dir:
+        output_paths.append(str(cert_dir))
+
+    manifest = build_run_manifest(
+        run_id=run_id,
+        config=config,
+        input_paths=[Path(candidate_path)]
+        + ([Path(reference_path)] if reference_path else [])
+        + [Path(config_path)],
+        output_paths=output_paths,
+        generated_at=generated_at,
+    )
+
+    if manifest_path:
+        write_json(manifest_path, manifest)
+    else:
+        manifest_out = Path(out_path).with_name("run_manifest.json")
+        write_json(manifest_out, manifest)
+
     return ranked
 
 
@@ -103,23 +190,27 @@ def write_report(
 ) -> None:
     p = Path(path)
     p.parent.mkdir(parents=True, exist_ok=True)
+    selected_ids = {item.candidate.candidate_id for item in selected}
     lines = [
-        "# OpenAMP Foundry Demo Report",
+        "# OpenAMP Foundry Candidate Report",
         "",
-        "This report is generated by transparent baseline heuristics.",
-        "It does not prove antimicrobial activity.",
+        "> **Disclaimer:** All scores are transparent baseline heuristics computed from "
+        "physicochemical properties. They are NOT validated biological predictors. "
+        "No antimicrobial activity has been demonstrated in vitro or in vivo. "
+        "These candidates are nominated for possible future expert review and assay only.",
         "",
         f"Total candidates scored: {len(ranked)}",
-        f"Candidates selected for evidence certificates: {len(selected)}",
+        f"Candidates selected (passed all filters, diverse): {len(selected)}",
         "",
-        "| Rank | ID | Sequence | Ensemble | Activity | Safety | Synthesis | Novelty |",
-        "|---:|---|---|---:|---:|---:|---:|---:|",
+        "| Rank | ID | Seq | Ensemble | Activity | Safety | Synthesis | Novelty | Selected |",
+        "|---:|---|---|---:|---:|---:|---:|---:|:---:|",
     ]
     for idx, item in enumerate(ranked, start=1):
         s = item.scores
+        sel_mark = "Y" if item.candidate.candidate_id in selected_ids else ""
         lines.append(
             f"| {idx} | {item.candidate.candidate_id} | `{item.candidate.sequence}` | "
             f"{s['ensemble']:.4f} | {s['activity']:.4f} | {s['safety']:.4f} | "
-            f"{s['synthesis']:.4f} | {s['novelty']:.4f} |"
+            f"{s['synthesis']:.4f} | {s['novelty']:.4f} | {sel_mark} |"
         )
     p.write_text("\n".join(lines) + "\n", encoding="utf-8")

src/openamp_foundry/types.py

@@ -20,3 +20,4 @@ class ScoredCandidate:
     nearest_reference: dict[str, Any] | None = None
     selection_reason: list[str] = field(default_factory=list)
     known_failure_modes: list[str] = field(default_factory=list)
+    valid: bool = True