nx_chem_adulterant_db.nx

nx_chem_adulterant_db.nx -- C4.0 milestone: first usable supplement-arc adulterant detection database + mass-window lookup pipeline. This is the smallest end-to-end demo: 1. Build seed DB of known supplement adulterants 2. Each entry parses real SMILES at DB-build time 3. Compute monoisotopic mass + [M+H]+ + halogen signature via C2.8a + C2.8b primitives 4. Lookup: given observed LC-MS peak m/z + ppm tolerance, scan DB and return matching entries 5. Report: human-readable lab-facing output with verdict Composes: C2.0 MolGraph + C2.1 SMILES parser + C2.3a implicit-H + C2.8a monoisotopic mass + ion m/z + C2.8b halogen sig. Seed compounds (publicly known supplement adulterants; FDA Tainted Supplements list + DSHEA + IOC banned list references): id 0 ephedrine C10H15NO restricted (DSHEA-pre-2004) id 1 DMAA C7H17N banned 2013 (FDA action) id 2 DMBA C6H15N banned 2014 (FDA action) id 3 DMHA C8H19N banned 2019 (FDA caution) id 4 sibutramine C17H26ClN banned 2010 (Abbott withdrew) id 5 caffeine C8H10N4O2 approved (baseline, not banned) id 6 higenamine C16H17NO3 banned 2017 (WADA + FDA) id 7 yohimbine C21H26N2O3 restricted (some markets) Honest gaps (deferred): - sildenafil (C22H30N6O4S; Viagra-active spiked in herbal ED): SMILES contains aromatic+saturated mixed ring system parser doesn't yet handle perfectly; C4.1. - Tadalafil (Cialis), anabolic steroid analogs, polyhalogenated: more complex parsing required; C4.1. - Persistent storage (memory-only DB at present); C4.2. - Multiple-adulterant interactions: not modeled; lab interpretation.