fix: address code review issues from PR #47 — charge model consistency, imports, defaults

- presynth_check.py: move selectivity_proxy imports to module level (was deferred;
  deferred imports are justified only to break circular dependencies, which do not
  exist here). Use physchem.net_charge_at_ph74() for consistency with compute_features()
  so the selectivity_proxy value is identical in both the feature dict and QC report.
  The old _net_charge_at_ph() included terminal groups/Cys/Tyr producing a ~0.2 offset
  that would give divergent cytotox verdicts for the same sequence.
- SynthQC.selectivity_proxy default changed from 1.0 to 0.0 (conservative: assume
  cytotoxic until check_sequence() fills the field; 1.0 = safe default is misleading
  for a risk field)
- physchem.py docstring: "optimum: +3 to +7" → "+2 to +7" (code awards full score
  starting at charge=2.0; the prior text contradicted the implementation and the test)
- test_selectivity_proxy.py: add boundary test (selectivity_proxy=0.50 exactly must NOT
  trigger cytotox_risk since predicate is strict <0.5)
- Makefile: update echo to "1081-test suite" → "1082-test suite" (38 new tests total)

Author: cschanhniem

Makefile

@@ -20,7 +20,7 @@ lint:
 	$(RUFF) check src tests scripts
 
 ci: lint test
-	@echo "CI passed: lint + 1044-test suite green"
+	@echo "CI passed: lint + 1081-test suite green"
 
 bench-leakage:
 	PYTHONPATH=src $(PYTHON) -m openamp_foundry.cli bench leakage \

src/openamp_foundry/features/physchem.py

@@ -145,7 +145,7 @@ def selectivity_proxy(charge: float, gravy: float) -> float:
     Based on Dathe & Wieprecht (1999) Biochim Biophys Acta 1462:71-87 (GRAVY/cytotox
     correlation) and Shai (2002) Biochim Biophys Acta 1462:55-70 (charge selectivity).
 
-    Selectivity optimum: net charge +3 to +7, GRAVY ≤ 0.5.
+    Selectivity optimum: net charge +2 to +7, GRAVY ≤ 0.5.
 
     - Charge < +2: insufficient electrostatic attraction to anionic bacterial membranes
       (phosphatidylglycerol, cardiolipin) → reduced bacterial killing, not cytotoxic.

src/openamp_foundry/qc/presynth_check.py

@@ -8,6 +8,10 @@
 import re
 from dataclasses import dataclass, field
 
+from openamp_foundry.features.physchem import net_charge_at_ph74 as _charge_at_ph74
+from openamp_foundry.features.physchem import selectivity_proxy as _sel_proxy
+from openamp_foundry.scoring.boman import gravy_score as _gravy
+
 
 # Trypsin cleaves after K or R (except when followed by P).
 _TRYPSIN_RE = re.compile(r"[KR](?!P)(?=.)")
@@ -124,8 +128,9 @@ class SynthQC:
     mu_h: float = 0.0
     hemolysis_start_conc: str = ""     # suggested starting concentration for hemolysis assay
 
-    # Mammalian cytotoxicity risk (selectivity proxy < 0.5)
-    selectivity_proxy: float = 1.0
+    # Mammalian cytotoxicity risk (selectivity proxy < 0.5).
+    # Default 0.0 = maximally conservative (assume cytotoxic until check_sequence() fills this).
+    selectivity_proxy: float = 0.0
     cytotox_risk: bool = False
 
     # C-terminal modification
@@ -261,12 +266,14 @@ def check_sequence(candidate_id: str, seq: str, mu_h: float = 0.0) -> SynthQC:
         qc.hemolysis_start_conc = f"Start at MIC; μH={mu_h:.2f} low risk"
 
     # Selectivity proxy (cytotoxicity risk).
-    # Computed from net charge and GRAVY to detect sequences likely to fail mammalian
-    # cytotoxicity assays. Flagged when proxy < 0.5 (high charge OR high hydrophobicity).
+    # Uses the same net_charge_at_ph74 model as compute_features() to guarantee that the
+    # selectivity_proxy value here matches the one stored in the candidate feature dict
+    # (side-chain only: K, R, H at pH 7.4 with pKa 6.5; D, E). The presynth_check
+    # _net_charge_at_ph() model includes terminal groups/Cys/Tyr, which produces a
+    # systematic offset (~0.2) and would cause divergent cytotox verdicts for the same
+    # sequence when comparing feature dicts to QC reports.
     # Literature: Dathe & Wieprecht (1999) BBA; Shai (2002) BBA; Chen et al. (2005) JBC.
-    from openamp_foundry.features.physchem import selectivity_proxy as _sel_proxy
-    from openamp_foundry.scoring.boman import gravy_score as _gravy
-    qc.selectivity_proxy = _sel_proxy(qc.charge_ph74, _gravy(seq))
+    qc.selectivity_proxy = _sel_proxy(_charge_at_ph74(seq), _gravy(seq))
     qc.cytotox_risk = qc.selectivity_proxy < 0.5
 
     # C-terminal amidation recommendation.

tests/test_selectivity_proxy.py

@@ -109,6 +109,13 @@ def test_very_hydrophobic_no_charge_is_zero(self):
     def test_return_type_is_float(self):
         assert isinstance(selectivity_proxy(5.0, 0.1), float)
 
+    def test_exact_threshold_boundary_is_not_flagged(self):
+        # selectivity_proxy(1.0, 0.75) = 0.6 × 0.5 + 0.4 × 0.5 = 0.50 exactly.
+        # The predicate is strict `< 0.5`, so 0.50 must NOT trigger HIGH_CYTOTOX_RISK.
+        p = selectivity_proxy(1.0, 0.75)
+        assert p == pytest.approx(0.50, abs=1e-4)
+        assert not (p < 0.5), "Boundary value 0.50 must NOT trigger cytotox risk (predicate is strict <)"
+
     def test_output_clipped_to_unit_interval(self):
         # Cannot exceed 1.0 or go below 0.0 for any input
         for charge in [-5, 0, 5, 15]: