Lab 5 — Vitamin D vehicle (Combs gate, not a DRI table)

Question for students: two supplements list the same vitamin D on the panel. One is an oil softgel. One is a dry tablet taken with water and no meal. Does the label IU arrive?

This lab extends Lab 2 from leaf carotenoids to cholecalciferol. Same constitution: gate ≠ bound, label ≠ dose, empty beats fake. It does not encode a requirement, a % absorbed, or a clinical protocol. Those stay in Combs / DRI / Krause.

The pair

from biology_as_code.packets import get_packet

for packet_id in ("ex.vitamin.d.softgel", "ex.vitamin.d.dry_tablet"):
    packet = get_packet(packet_id)
    print(packet.common_name)
    print("  cargo :", packet.cargo_nutrients())
    print("  lipid :", packet.partner("dietary_lipid_g"), "g")
    print("  status:", packet.status if hasattr(packet, "status") else "filled")

Same named cargo (cholecalciferol). One declares a lipid partner, the other declares 0. Label milligrams / IU stay open on purpose.

Audit a disease slogan against both

from biology_as_code import Claim, audit_claim
from biology_as_code.packets import get_packet

claim = Claim(
    id="claim.vitd_prevents_rickets",
    surface_claim="Take this vitamin D tablet to prevent rickets",
    verb_class="disease_claim",
    nutrient="cholecalciferol",
    surface_verb="prevents",
)

for packet_id in ("ex.vitamin.d.dry_tablet", "ex.vitamin.d.softgel"):
    result = audit_claim(claim, get_packet(packet_id))
    print(f"\n--- {packet_id}")
    print("verdict        :", result.verdict)
    print("gate_check     :", result.gate_check)
    print("closed_through :", result.l1_to_l5.get("closed_through"))
    print("laws           :", result.law_refs)

Expected teaching shape (not a clinical answer):

• Dry tablet: mechanism gate fails at L3 if the auditor treats cholecalciferol like other fat-soluble cargo (lipid required). That is a closed path, not a smaller number.
• Softgel: lipid gate can pass, but a disease verb must still die at L5 (UNEVALUABLE). One capsule is not a rickets trial.

If cholecalciferol is not yet on the fat-soluble gate register, the honest result is UNEVALUABLE — not a invented pass. Do not patch a green score over a missing nutrient alias. File that as a follow-up law-register PR.

Weaker verb

mechanism_claim = Claim(
    id="claim.vitd_needs_lipid",
    surface_claim="An oil vehicle makes this vitamin D absorbable via micelles",
    verb_class="bound_increase",
    nutrient="cholecalciferol",
)
result = audit_claim(mechanism_claim, get_packet("ex.vitamin.d.softgel"))
print(result.verdict, result.gate_check)

A mechanism claim may be Plausible when the vehicle is declared. A rickets claim is not. The packet did not change into a textbook chapter on DRI or pediatrics.

What this lab refuses to compute

• IU → nmol/L 25(OH)D
• Host adiposity, malabsorption, or latitude
• Infant drops vs adult capsules as clinical protocols

Those are Brown / Krause / Combs chapters. Code here only asks: was the lipid seat occupied?

Exercise

1. Compare ex.vitamin.d.dry_tablet to ex.spinach_salad.zero_fat. Same cliff, different cargo. Name the cargo and the shared gate field.
2. A dry tablet swallowed with a 15 g fat meal — which seat moved, packet or meal partner? Should that be a new packet or a composed meal object?
3. If gates_for("cholecalciferol") is empty, is the correct next commit a fake % absorption table, or a LAW-020 alias row? Argue from the constitution.