factorforge-cds 3.2.7__tar.gz → 3.3.0__tar.gz

{factorforge_cds-3.2.7/src/factorforge_cds.egg-info → factorforge_cds-3.3.0}/PKG-INFO

@@ -1,7 +1,7 @@
 Metadata-Version: 2.4
 Name: factorforge-cds
-Version: 3.2.7
-Summary: FactorForge - open-source CDS design and pre-synthesis sequence review engine by Eijex.
+Version: 3.3.0
+Summary: FactorForge - open-source CDS design and pre-synthesis sequence review by Eijex.
 Author-email: Eijex <eijex.lab@gmail.com>
 License-Expression: AGPL-3.0-only
 Project-URL: Homepage, https://factorforge.eijex.com
@@ -33,7 +33,7 @@ Dynamic: license-file
 
 # FactorForge
 
-**Open-source constraint-based CDS design and pre-synthesis sequence review engine for plant CDS workflows, with primary support for *Nicotiana benthamiana* (Tobacco BY-2: experimental).**
+**Open-source constraint-based CDS design and pre-synthesis sequence review for plant CDS workflows, with primary support for *Nicotiana benthamiana* (Tobacco BY-2: experimental).**
 
 [![License](https://img.shields.io/badge/license-AGPL--3.0-blue.svg)](LICENSE)
 [![Python](https://img.shields.io/badge/python-3.10%2B-blue.svg)](https://www.python.org/)
@@ -93,7 +93,7 @@ FactorForge outputs are **in-silico only** and have not been experimentally vali
 ## Citing
 
 ```
-FactorForge v3.2.7 (2026). Open-source constraint-based CDS design engine.
+FactorForge v3.3.0 (2026). Open-source constraint-based CDS design and sequence review.
 Eijex. https://github.com/eijex/factorforge-cds
 ```
 

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/README.md

@@ -1,6 +1,6 @@
 # FactorForge
 
-**Open-source constraint-based CDS design and pre-synthesis sequence review engine for plant CDS workflows, with primary support for *Nicotiana benthamiana* (Tobacco BY-2: experimental).**
+**Open-source constraint-based CDS design and pre-synthesis sequence review for plant CDS workflows, with primary support for *Nicotiana benthamiana* (Tobacco BY-2: experimental).**
 
 [![License](https://img.shields.io/badge/license-AGPL--3.0-blue.svg)](LICENSE)
 [![Python](https://img.shields.io/badge/python-3.10%2B-blue.svg)](https://www.python.org/)
@@ -60,7 +60,7 @@ FactorForge outputs are **in-silico only** and have not been experimentally vali
 ## Citing
 
 ```
-FactorForge v3.2.7 (2026). Open-source constraint-based CDS design engine.
+FactorForge v3.3.0 (2026). Open-source constraint-based CDS design and sequence review.
 Eijex. https://github.com/eijex/factorforge-cds
 ```
 

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/pyproject.toml

@@ -4,8 +4,8 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "factorforge-cds"
-version = "3.2.7"
-description = "FactorForge - open-source CDS design and pre-synthesis sequence review engine by Eijex."
+version = "3.3.0"
+description = "FactorForge - open-source CDS design and pre-synthesis sequence review by Eijex."
 readme = "README.md"
 license = "AGPL-3.0-only"
 license-files = ["LICENSE"]

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/src/factorforge/__init__.py

@@ -4,7 +4,7 @@ FactorForge - Codon Optimization Platform
 profile: constraint-aware rule/profile engine
 """
 
-__version__ = "3.2.7"
+__version__ = "3.3.0"
 __author__ = "Eijex"
 
 # Auto-register engines (safe when running from source tree)

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/src/factorforge/analysis/feasibility.py

@@ -14,13 +14,11 @@ from factorforge.analysis.metrics import (
 )
 
 
-# DEFAULT_CAI_TARGET=0.82 aligns with industry practice (>0.8) and is achievable
-# (internal benchmark, n=49, avg CAI=0.76).
-# DEFAULT_GC_LOW/HIGH = native genome-composition anchor for N. benthamiana CDS
-# (_analysis/025 STEP 2: 004 endogenous n=10 measured range 40-47%, cross-checked
-# against nbev11_cds_hc/all and qld183_v103 derived-asset GC ~42.8-43.1% and
-# external ground truth ~44%). NOT an empirically validated expression optimum —
-# this is a composition anchor, not a target to maximize toward.
+# Defaults for the current N. benthamiana software-default codon reference.
+# DEFAULT_CAI_TARGET=0.82 remains a soft in-silico target aligned with
+# industry practice (>0.8). DEFAULT_GC_LOW/HIGH track the NbeV1.1
+# high-confidence CDS-derived native composition band. These are not wet-lab
+# validation, expression/yield prediction, or biological-superiority claims.
 # Exported as named constants so tests/test_registry_production_sync.py can
 # strictly compare them against the registry (single source of truth).
 DEFAULT_CAI_TARGET: float = 0.82
@@ -120,11 +118,7 @@ def analyze_feasibility(
     if not protein:
         raise ValueError("protein_sequence must not be empty")
 
-    # Default exploration ranges: genome-grounded native anchor first, then
-    # progressively wider windows. (55.0, 65.0) is retained ONLY as an explicit
-    # non-native/high-GC option (matches legacy engine-output-calibrated band) —
-    # callers must opt in explicitly to it, it is not the production default.
-    ranges = gc_ranges or [(40.0, 47.0), (35.0, 50.0), (55.0, 65.0)]
+    ranges = gc_ranges or [(40.0, 47.0), (40.0, 55.0), (40.0, 65.0)]
     normalized_ranges = [
         (_normalize_gc_bound(low), _normalize_gc_bound(high)) for low, high in ranges
     ]

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/src/factorforge/analysis/metrics.py

@@ -98,7 +98,7 @@ class CodonUsageTable:
 
 
 def _default_codon_table_path() -> Path:
-    # Job 168 / v3.3.0 (_analysis/025): production default switched from the
+    # v3.3.0 reference-policy update: production default switched from the
     # legacy Kazusa/SGN-derived table to the NbeV1.1 LAB-strain high-confidence
     # derived table. See data/reference/active_codon_reference.json.
     return get_data_path() / "profiles" / "nbev11_cds_hc_derived_codons.json"

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/src/factorforge/engines/__init__.py

@@ -13,7 +13,7 @@ def register_builtin_engines() -> None:
         "profile",
         RuleBasedOptimizer,
         metadata={
-            "version": "3.2.7",
+            "version": "3.3.0",
             "engine_type": "profile_rule_based",
             "role": "stable_profile_engine",
             "stable": True,

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/src/factorforge/engines/profile/__init__.py

@@ -5,7 +5,7 @@ Production system (2026)
 Plant-specific rule-based optimization
 """
 
-__version__ = "3.2.7"
+__version__ = "3.3.0"
 
 from .optimizer import RuleBasedOptimizer
 from .pipeline import OptimizationPipeline

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/src/factorforge/engines/profile/optimizer.py

@@ -10,7 +10,7 @@ from factorforge.core.interfaces import OptimizationResult, OptimizerEngine
 from .exporter import SequenceExporter
 from .rules.reverse_translator import OptimizationProfile, ReverseTranslator
 from .rules.rule_engine import RuleEngine
-from .scoring import calculate_composite_score, compute_mfe_evidence
+from .scoring import calculate_composite_score, compute_mfe_evidence, resolve_host_gc_range
 from .validator import InputValidator
 
 logger = logging.getLogger(__name__)
@@ -20,7 +20,7 @@ class RuleBasedOptimizer(OptimizerEngine):
     """Profile-based rule optimization engine."""
 
     name = "Profile-based"
-    version = "3.2.7"
+    version = "3.3.0"
 
     def __init__(self, codon_table_path: str | None = None) -> None:
         """
@@ -157,6 +157,22 @@ class RuleBasedOptimizer(OptimizerEngine):
             "score": candidates[0]["score"],
             "violations": sum(len(v) for v in scan_results.values()),
         }
+        if profile_value == "balanced":
+            host_gc_min, host_gc_max = resolve_host_gc_range(host)
+            requested_gc_min_percent = float(kwargs.get("target_gc_min", host_gc_min))
+            requested_gc_max_percent = float(kwargs.get("target_gc_max", host_gc_max))
+            achieved_gc_percent = metrics["gc_percent"]
+            metrics.update(
+                {
+                    "gc_target_reached": (
+                        requested_gc_min_percent
+                        <= achieved_gc_percent
+                        <= requested_gc_max_percent
+                    ),
+                    "requested_gc_min_percent": requested_gc_min_percent,
+                    "requested_gc_max_percent": requested_gc_max_percent,
+                }
+            )
         # MFE provenance: expose whether MFE was actually computed so downstream
         # artifacts (API response, Design Package) never report an uncomputed
         # MFE as a misleading 0.0 (016 audit). Score value is unchanged.

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/src/factorforge/engines/profile/rules/reverse_translator.py

@@ -374,11 +374,14 @@ class ReverseTranslator:
         Balanced profile: CAI first, GC balanced
 
         - Preferred codon ratio: 70%
-        - Target GC: host composition band (nbenthamiana: 40-47%, native
-          genome-composition anchor; see _analysis/025 STEP 2 and
-          engines/profile/scoring.py GC_OPT_MIN/MAX. Not an empirically
-          validated expression optimum. Other hosts: resolve_host_gc_range()
-          default, unchanged from pre-v3.3.0 behavior.)
+        - Target GC: host composition band, resolved via
+          resolve_host_gc_range() (engines/profile/scoring.py
+          GC_OPT_MIN/MAX). N. benthamiana currently uses the NbeV1.1
+          high-confidence CDS-derived native-composition band (40-47%);
+          other hosts retain their configured defaults.
+        - GC target requests are best-effort sampling targets, not guarantees:
+          achieved GC can be dominated by the codon table's frequency-weighted
+          sampling distribution.
         """
         _host_gc_min, _host_gc_max = resolve_host_gc_range(self.host)
         target_gc_min = kwargs.get("target_gc_min", _host_gc_min)
@@ -453,7 +456,7 @@ class ReverseTranslator:
 
         Targets the caller-supplied ``target_gc`` if provided, otherwise the
         active host's composition midpoint (resolve_host_gc_range(self.host);
-        43.5% for nbenthamiana since v3.3.0 / _analysis/025, 60.0% for other
+        43.5% for nbenthamiana since v3.3.0 / reference-policy audit, 60.0% for other
         hosts pending their own host-specific analysis). To target a different
         GC (e.g. for specific vector requirements), pass target_gc explicitly.
 

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/src/factorforge/engines/profile/rules/rule_engine.py

@@ -358,12 +358,14 @@ class RuleEngine:
 
         This is a LOCAL synthesis/extreme-window guard (default 25-75% over a
         50 bp window), NOT the global GC target. Global GC is governed separately
-        by the scoring band (GC_OPT_MIN/MAX, native genome-composition anchor
-        ~40-47%, see _analysis/025 STEP 2) and the API/DP gc_min/gc_max
-        constraints. The wide 25-75% local band is an independent synthesis/
-        hairpin-risk guard (registry `gc_extreme_local_window`), not derived from
-        or coupled to the global GC target — narrowing it is a separate decision
-        from the global band retarget and is out of scope here.
+        by the scoring band (GC_OPT_MIN/MAX; N. benthamiana currently uses
+        the NbeV1.1 high-confidence CDS-derived native-composition anchor,
+        40-47%, while other hosts retain their own configured defaults) and
+        the API/DP gc_min/gc_max constraints. The wide 25-75% local band is
+        an independent synthesis/hairpin-risk guard
+        (registry `gc_extreme_local_window`), not derived from or coupled to
+        the global GC target — narrowing it is a separate decision from the
+        global band retarget and is out of scope here.
 
         Args:
             seq: DNA sequence

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/src/factorforge/engines/profile/scoring.py

@@ -11,23 +11,20 @@ from typing import Any
 
 logger = logging.getLogger(__name__)
 
-# GC band for N. benthamiana codon-optimized sequences.
-# Native genome-composition anchor (_analysis/025 STEP 2: 004 endogenous CDS
-# n=10 measured range 40-47%; cross-checked against nbev11_cds_hc/all and
-# qld183_v103 derived-asset GC ~42.8-43.1% and external ground truth ~44%).
-# NOT an empirically validated expression optimum — this is a composition
-# anchor, not a target to maximize toward. Sequences within [GC_OPT_MIN,
-# GC_OPT_MAX] receive full GC score; outside the band the score decays linearly.
+# GC band for the current N. benthamiana software-default codon reference.
+# NbeV1.1 high-confidence CDS-derived codon usage anchors the public default
+# to the native genome-derived composition band. This is a configured
+# in-silico reference/GC policy, not wet-lab validation or expression/yield
+# prediction. Sequences inside [GC_OPT_MIN, GC_OPT_MAX] receive full GC score;
+# outside the band the score decays linearly.
 GC_OPT_MIN = 40.0
 GC_OPT_MAX = 47.0
 GC_OPT_MID = 43.5  # kept for gc_target point-scoring and viral_delivery centering
 GC_DECAY_WIDTH = 20.0  # percentage points outside band before score reaches 0.0
 
-# Job 168 / v3.3.0 (_analysis/025) scoped the genome-composition re-derivation
-# to N. benthamiana only. Other hosts (e.g. ntabacum/BY-2) keep the pre-v3.3.0
-# global default (55-65%, internal benchmark n=49 avg GC=60.1%) until they get
-# their own host-specific genome-composition analysis — they must NOT silently
-# inherit GC_OPT_MIN/MAX, which is an N.-benthamiana-specific anchor.
+# Host-isolation fix: ntabacum/BY-2 must not silently inherit the NbeV1.1
+# N. benthamiana band. Other hosts keep the pre-v3.3.0 global default until
+# they get their own host-specific analysis.
 GC_RANGE_DEFAULT: tuple[float, float] = (55.0, 65.0)
 GC_RANGES_BY_HOST: dict[str, tuple[float, float]] = {
     "nbenthamiana": (GC_OPT_MIN, GC_OPT_MAX),
@@ -37,7 +34,7 @@ GC_RANGES_BY_HOST: dict[str, tuple[float, float]] = {
 def resolve_host_gc_range(host: str | None) -> tuple[float, float]:
     """Resolve the (gc_min, gc_max) composition band for a host.
 
-    nbenthamiana uses the _analysis/025 native genome-composition anchor.
+    nbenthamiana uses the reference-policy audit native genome-composition anchor.
     Any other host (including unknown ones) keeps the pre-v3.3.0 global
     default band until it gets its own host-specific analysis.
     """
@@ -142,7 +139,7 @@ def _check_vienna_available() -> bool:
     return _vienna_available
 
 
-# 170-fix: ViennaRNA's RNA.fold() uses Zuker's MFE algorithm, O(n^3) time /
+# ViennaRNA's RNA.fold() uses Zuker's MFE algorithm, O(n^3) time /
 # O(n^2) memory — there was previously no length guard anywhere in the
 # calculate_mfe() call chain, so a single request at/under the existing
 # public API length limits (5000aa/15000bp) could pin a CPU core for minutes
@@ -175,7 +172,7 @@ def calculate_mfe(sequence: str) -> float | None:
             "Sequence length (%d nt) exceeds MFE_MAX_SEQUENCE_LENGTH (%d nt); "
             "skipping global MFE calculation to avoid an unbounded ViennaRNA "
             "RNA.fold() runtime (O(n^3)). MFE scoring falls back to a neutral, "
-            "zero-weighted contribution for this candidate (170-fix).",
+            "zero-weighted contribution for this candidate.",
             len(sequence), MFE_MAX_SEQUENCE_LENGTH,
         )
         return None
@@ -196,8 +193,7 @@ def normalize_mfe(mfe: float, seq_length: int) -> float:
     """
     Normalize MFE to 0-1 range where 1 = no structure (favorable).
 
-    Clamp range calibrated empirically in analysis 011
-    (eijex-workspace/_analysis/2026-06-26/011-mfe-clamp-calibration):
+    Clamp range calibrated empirically from internal computational audit data:
     measured MFE/nt across 135 FactorForge outputs (N. benthamiana +
     BY-2 hosts, 5 profiles) ranged -0.4064 to -0.1338 (combined 5th/95th
     percentile -0.3839/-0.1760). The range below widens that empirical
@@ -211,8 +207,7 @@ def normalize_mfe(mfe: float, seq_length: int) -> float:
     Returns:
         Normalized MFE score (0-1) under this computational normalization,
         where a higher value represents a less negative whole-CDS MFE/nt.
-        This is a Tier-0 computational heuristic only (see
-        eijex-validationHub/docs/CLAIM_EVIDENCE_BENCHMARK_MODEL.md §4) — no
+        This is a Tier-0 computational heuristic only — no
         biological interpretation (translation efficiency, mRNA stability,
         or expression outcome) is implied or validated.
     """
@@ -418,7 +413,7 @@ def compute_mfe_evidence(
     elif not _check_vienna_available():
         reason = "MFE was not computed because ViennaRNA is unavailable in this environment."
     elif len(sequence) > MFE_MAX_SEQUENCE_LENGTH:
-        # 170-fix: distinguish a deliberate length-based skip from an actual
+        # Distinguish a deliberate length-based skip from an actual
         # fold failure — the generic "computation failed" message below would
         # otherwise mislead a caller into thinking something is broken.
         reason = (

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/src/factorforge/engines/profile/utils.py

@@ -115,11 +115,10 @@ def calculate_dinucleotide_ratio(sequence: str, dinucleotide: str = "CG") -> flo
     return observed / expected
 
 
-# Job 168 / v3.3.0 (_analysis/025): host -> production-default codon table file
-# overrides. nbenthamiana moved from the legacy {host}_codons.json convention to
-# the NbeV1.1 LAB-strain derived table. See data/reference/active_codon_reference.json.
-# The legacy file itself (nbenthamiana_codons.json) is left untouched on disk for
-# historical benchmark replay (benchmarks/run_benchmark.py --codon-table-path).
+# NbeV1.1 high-confidence CDS-derived codon usage is the production-default
+# N. benthamiana software reference. This is a provenance-tracked in-silico
+# default, not wet-lab validation or a biological-superiority claim; see
+# data/reference/active_codon_reference.json.
 _HOST_CODON_TABLE_OVERRIDES: dict[str, str] = {
     "nbenthamiana": "profiles/nbev11_cds_hc_derived_codons.json",
 }

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0/src/factorforge_cds.egg-info}/PKG-INFO

@@ -1,7 +1,7 @@
 Metadata-Version: 2.4
 Name: factorforge-cds
-Version: 3.2.7
-Summary: FactorForge - open-source CDS design and pre-synthesis sequence review engine by Eijex.
+Version: 3.3.0
+Summary: FactorForge - open-source CDS design and pre-synthesis sequence review by Eijex.
 Author-email: Eijex <eijex.lab@gmail.com>
 License-Expression: AGPL-3.0-only
 Project-URL: Homepage, https://factorforge.eijex.com
@@ -33,7 +33,7 @@ Dynamic: license-file
 
 # FactorForge
 
-**Open-source constraint-based CDS design and pre-synthesis sequence review engine for plant CDS workflows, with primary support for *Nicotiana benthamiana* (Tobacco BY-2: experimental).**
+**Open-source constraint-based CDS design and pre-synthesis sequence review for plant CDS workflows, with primary support for *Nicotiana benthamiana* (Tobacco BY-2: experimental).**
 
 [![License](https://img.shields.io/badge/license-AGPL--3.0-blue.svg)](LICENSE)
 [![Python](https://img.shields.io/badge/python-3.10%2B-blue.svg)](https://www.python.org/)
@@ -93,7 +93,7 @@ FactorForge outputs are **in-silico only** and have not been experimentally vali
 ## Citing
 
 ```
-FactorForge v3.2.7 (2026). Open-source constraint-based CDS design engine.
+FactorForge v3.3.0 (2026). Open-source constraint-based CDS design and sequence review.
 Eijex. https://github.com/eijex/factorforge-cds
 ```
 

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/tests/test_benchmark_codon_table_metadata.py

@@ -111,8 +111,8 @@ def test_smoke_summary_contains_codon_table_fields(tmp_path):
         assert field in data, f"Missing codon table field in smoke summary JSON: {field}"
 
     # No --codon-table-path override ⇒ codon_table_id must reflect whichever
-    # asset is actually the current production default (Job 168 / v3.3.0,
-    # _analysis/025 — see data/reference/active_codon_reference.json), not a
+    # asset is actually the current production default (v3.3.0 reference-policy update,
+    # reference-policy audit — see data/reference/active_codon_reference.json), not a
     # hardcoded legacy literal.
     active_ref = json.loads(
         (ROOT / "data" / "reference" / "active_codon_reference.json").read_text(encoding="utf-8")
@@ -148,9 +148,6 @@ def test_smoke_summary_contains_vienna_rna_active(tmp_path):
 
 def test_no_raw_genome_fasta_committed():
     """Raw genome/CDS/protein FASTA files must not be committed to the repo."""
-    forbidden_patterns = ["*.fasta", "*.fa", "*.fna", "*.fastq"]
-    allowed_fixtures = ROOT / "tests" / "fixtures"
-
     import subprocess
     result = subprocess.run(
         ["git", "ls-files", "--", "*.fasta", "*.fa", "*.fna", "*.fastq"],
@@ -168,6 +165,6 @@ def test_no_raw_genome_fasta_committed():
         if p and not any(p.startswith(pfx) for pfx in allowed_prefixes)
     ]
     assert not committed, (
-        f"Raw FASTA/genome files must not be committed outside allowed locations:\n"
+        "Raw FASTA/genome files must not be committed outside allowed locations:\n"
         + "\n".join(committed)
     )

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/tests/test_registry_production_sync.py

@@ -76,7 +76,7 @@ def test_aa_identity_policy_sync():
     assert ok["passed"] and ok["aa_identity"] == registry_val
 
 
-# ── codon_reference source-of-truth sync (Job 168 / v3.3.0, _analysis/025) ────
+# ── codon_reference source-of-truth sync (v3.3.0 reference-policy update) ────
 
 def test_codon_reference_active_sync_with_active_reference_file():
     """registry's codon_reference.active block must match
@@ -105,7 +105,7 @@ def test_codon_reference_active_sync_with_active_reference_file():
 
 def test_codon_reference_active_sync_with_v2_manifest():
     """registry's codon_reference.active block must match the schema-conformant
-    v2 manifest file's facts (asset_type, sha256, source_status)."""
+    NbeV1.1 HC (v2) manifest file's facts now that v2 is the production default."""
     import json
     from pathlib import Path
 
@@ -120,6 +120,23 @@ def test_codon_reference_active_sync_with_v2_manifest():
     assert registry_active["source_status"] == manifest["source_status"]
 
 
+def test_codon_reference_historical_comparator_sync_with_legacy_manifest():
+    """registry's codon_reference.historical_comparator block must stay in sync
+    with the legacy v1 manifest retained for v3.2.x continuity/replay."""
+    import json
+    from pathlib import Path
+
+    registry_historical = _resolve("codon_reference.historical_comparator")
+    manifest = json.loads(
+        (Path(__file__).resolve().parents[1] / "data" / "reference"
+         / "codon_table_manifest.json").read_text(encoding="utf-8")
+    )
+    assert registry_historical["id"] == manifest["codon_table_id"]
+    assert registry_historical["sha256"] == manifest["sha256"]
+    assert registry_historical["asset_type"] == manifest["asset_type"]
+    assert registry_historical["source_status"] == manifest["source_status"]
+
+
 def test_codon_reference_active_table_sha256_matches_production_default():
     """The sha256 recorded for the active codon_reference must match the
     actual file the production engine resolves to by default."""

{factorforge_cds-3.2.7 → factorforge_cds-3.3.0}/tests/test_worked_example.py

@@ -63,7 +63,7 @@ def test_sfgfp_sequence_no_x_residues():
 def test_run_example_deterministic():
     """run_example.py (no --freeze) must exit 0 when frozen outputs exist.
 
-    Job 168 / v3.3.0 (_analysis/025): this script is explicitly pinned to the
+    v3.3.0 reference-policy update: this script is explicitly pinned to the
     legacy v1 codon reference (GC 55-65%) regardless of the engine's current
     production default, so this reproducibility check stays valid forever.
     See run_example_v2_smoke.py / test_run_example_v2_smoke_succeeds for the
@@ -82,10 +82,13 @@ def test_run_example_deterministic():
 
 
 def test_run_example_v2_smoke_succeeds():
-    """run_example_v2_smoke.py (current production default) must exit 0.
+    """run_example_v2_smoke.py (tracks whatever the current production
+    default actually is) must exit 0 and report a provenance ID that matches
+    data/reference/active_codon_reference.json — not hardcoded to v2, since
+    the example is aligned to the current v3.3.0 production-default reference policy.
 
-    No frozen-output comparison — this only checks that the v2 path runs
-    end-to-end and reports correct provenance (Job 168 / v3.3.0, _analysis/025).
+    No frozen-output comparison — this only checks that the current-default
+    path runs end-to-end and reports correct provenance.
     """
     result = subprocess.run(
         [sys.executable, str(EXAMPLE_DIR / "run_example_v2_smoke.py")],
@@ -98,7 +101,8 @@ def test_run_example_v2_smoke_succeeds():
         f"stdout:{result.stdout}\nstderr:{result.stderr}"
     )
     assert "OK" in result.stdout
-    assert "nbenthamiana_nbev11_hc_v2" in result.stdout
+    active_ref = _load(ROOT / "data" / "reference" / "active_codon_reference.json")
+    assert active_ref["active_codon_table_id"] in result.stdout
 
 
 # ---------------------------------------------------------------------------