Systematic Under-Characterization of Psychedelic Compound Classes in Published hERG QSAR Training Data: A Pre-Registered Multi-Family Analysis

Abstract

Pre-registered multi-family hERG QSAR validity analysis across three structurally independent psychedelic compound families (22 tryptamines, 20 phenethylamines, 17 cathinones; 59 compounds total). Central thesis (pre-registered): psychedelic-class compounds are systematically under-characterized in published primary hERG patch-clamp literature with same-preparation Ki + IC50 pairing under operational definition v2. Phase 1 indexed-database retrieval (ChEMBL CHEMBL240 + DrugMatrix AID 625171 + BindingDB UniProt Q12809) returned a single substantive datapoint across 59 compounds: Bupropion patch-clamp IC50 = 69 microM (weak blocker, CHEMBL5726960). A pre-registered targeted full-text retrieval protocol (Section 6) tests whether under-characterization extends to the published-literature surface (central thesis CONFIRMED at 0-4 missed pairs) or reflects an indexed-database-only gap (central thesis WEAKENED at >=5 stratified-inclusion missed pairs). Tiered disposition outcomes pre-committed (Section 1.1) across stratified-inclusion default tier and strict-recording-method stronger-weakening tier. Pre-registered priors falsification cluster (6 of 11 falsified on the meta-assumption "well-known compound classes / marketed pharma / regulatory-pipeline compounds have published primary patch-clamp data") is itself empirical evidence for the central thesis. Second co-equal contribution (methodological): registers a third pre-registration template family for chemoinformatics QSAR-validity work, the data-landscape-primary template (Section 6.3), structurally distinct from per-compound-prediction templates (Paper I) and predictive-model-specification templates (originally-scoped Paper II under predictive-theory framing). Template 3 pre-registers a search protocol as load-bearing methodology, ties central-thesis disposition to retrieval outcome, structurally frames secondary contributions as evidence-for the primary finding, and pre-registers conditional recommendations per disposition outcome. Six-item transferability checklist (Section 6.5) operationalizes the template for adoption by other groups. Methodology imported from Substrate Geometry research program (pre-commit + falsification + amendment-of-record + audit-trail-hash-chain discipline; Couey 2026, methodology manuscript in arXiv submission queue). Companion paper: Couey 2026 Paper I (architecture-specific failure modes in hERG QSAR on iboga alkaloids; ChemRxiv preprint + OSF preregistration uwvx4). Pre-registration document SHA-256 (markdown source): E6EA40702C74EEF8EEE24C22514BF10FF51791B4F4D13EBD5D2E0DADD44EE152. Archival PDF render SHA-256: FC4F7B26D86ACFD42EF96D2393EF04AF9F582CA53DFB5DCD1F003F955D6D09BB. Audit trail JSON SHA-256 at registration: 79DD717B915BD2C3D77F6BB515B067278BE79D90BAE5A52E3561C5B7707B5DB4 (26 anchors through integration-review-fixes entry). Sequencing discipline (per pre-registration Section 6): targeted full-text retrieval (Section 6.1 bath-salt forensic toxicology + Section 6.2 Bupropion FDA NDA filings + Section 6.3 ergoline boundary spot-check) and three-architecture canonical-SMILES prediction pass on 59 compounds (Pred-hERG 5.0 + ADMET-AI 2.0.1 + admetSAR 2.0) execute post-OSF-publish per pre-registration timestamp anchor. Composition Results data-filling and ChemRxiv preprint submission follow.

Topics & conditions

Detected automatically by Psychedelic Commons from the title and abstract, not labels from the source. They may be incomplete.

Keywords