ChEMBL Cross-Check — Use Boundary and Source-Verification Workflow¶
Compound refresh table¶
Last refreshed: 2026-10-01. All compounds re-queried via the ChEMBL REST API (https://www.ebi.ac.uk/chembl/api/data/); ChEMBL version was not returned by the API in this run (see logs/chembl-refresh-state.json). Entries marked ★ are new to this quarter's sweep.
Evidence level for all ChEMBL bioactivity records defaults to (In Vitro) unless the underlying reference is explicitly clinical. "pChEMBL" is −log₁₀(molar IC50/Ki/EC50); higher = more potent. Top-target column shows the highest-pChEMBL named protein target, not the most biologically relevant one — see the per-record verification rules below before using any value in a wiki claim.
Indexed compounds¶
| Compound | ChEMBL ID | Top annotated target | Best pChEMBL | Notes | Last refreshed |
|---|---|---|---|---|---|
| Dapansutrile | CHEMBL3989943 | NLRP3 (IC50) | 9.00 | MoA confirmed (direct inhibitor); max_phase 4 | 2026-10-01 |
| Colchicine | CHEMBL107 | Tubulin / beta-tubulin binding (MoA) | — | Anti-mitotic; cell-line phenotypic hits dominate top records | 2026-10-01 |
| Zileuton | CHEMBL93 | ALOX5 / 5-LOX (IC50) | 6.85 | MoA confirmed; max_phase 4 | 2026-10-01 |
| Oridonin | CHEMBL1164920 | NLRP3 (IC50) | 6.11 | 408 records; covalent NEK7/NLRP3 mechanism | 2026-10-01 |
| Ursolic acid | CHEMBL169 | ROR-gamma (IC50) | 9.12 | Prior queue item (2026-Q3) resolved; NF-kB p65 also annotated (7.51) | 2026-10-01 |
| Cholecalciferol | CHEMBL1042 | VDR (EC50) | 9.68 | Active metabolite calcitriol (CHEMBL846) has richer dataset | 2026-10-01 |
| Genistein | CHEMBL44 | ERβ (Ki) | 9.23 | 2064 records; also tyrosine kinase and CFTR | 2026-10-01 |
| Curcumin | CHEMBL140 | Amyloid-beta APP (Ki 9.68); DYRK2 (IC50 8.60) | 9.68 | PAINS compound; prior DYRK2 queue item (2026-Q3) resolved | 2026-10-01 |
| Quercetin | CHEMBL50 | MAO-A (IC50) | 8.00 | 2978 records; very broad activity profile | 2026-10-01 |
| EGCG | CHEMBL297453 | Bacterial FabI (Ki) | 8.10 | Top hits are antimicrobial; pref_name typo in ChEMBL ("EPIGALOCATECHIN") | 2026-10-01 |
| Disulfiram | CHEMBL964 | LOXL4 (IC50); ALOX15 (screen) | 7.23 | GSDMD-pore mechanism (cited MoA) not in top-5 pChEMBL records | 2026-10-01 |
| Tranilast | CHEMBL415324 | NPC1 / Rab-9A (NCATS screen) | 7.05 | No formal MoA in ChEMBL mechanism table | 2026-10-01 |
| NAC | CHEMBL600 | ALOX15 (NCATS screen) | 7.60 | Indexed as "Acetylcysteine"; most records use non-pChEMBL readouts | 2026-10-01 |
| Cordycepin | CHEMBL305686 | Antiparasitic / antiviral (functional) | 9.30 | Top-pChEMBL records are anti-trypanosomal; adenosine kinase in broader set | 2026-10-01 |
| Spermidine | CHEMBL19612 | Carbonic anhydrase (Ki) | 6.95 | Autophagy / mTOR mechanism not captured in ChEMBL records | 2026-10-01 |
| Carnosine | CHEMBL242948 | PEPT2 / SLC15A2 (EC50, substrate) | 5.65 | Dipeptide transporter substrate; ROR-gamma screen potency 5.65 (one document) | 2026-10-01 |
| Sulforaphane | CHEMBL48802 | iNOS (IC50); Nrf2 / Keap1 (EC50) | 6.40 | No direct NLRP3 binding in top records | 2026-10-01 |
| Beta-caryophyllene | CHEMBL445740 | CB2 (Ki) | 6.82 | 8 pChEMBL records total; sparse | 2026-10-01 |
| Salidroside ★ | CHEMBL465208 | MAO-B (IC50) | 6.09 | NEW; MSU animal lead (PMID 30265377) vs MAO-B ChEMBL top target — see discrepancy block | 2026-10-01 |
| BHB | CHEMBL1162496 | MCT1 / MCT2 (transport assays) | — | Indexed as "(+/−)-3-hydroxybutyric acid"; 0 pChEMBL records | 2026-10-01 |
| Theaflavin (TF1) | CHEMBL346119 | Bcl-2 (Ki) | 6.16 | 5 pChEMBL records; separate entry for digallate | 2026-10-01 |
| Theaflavin-3,3'-digallate (TF3) | CHEMBL434864 | OATP1B1 (Ki); OATP1B3 (Ki) | 5.82 | Hepatic uptake transporters, not renal; see discrepancy block | 2026-10-01 |
| Eurycomanone | CHEMBL1171981 | Unchecked (IC50, sparse) | 5.62 | Active constituent of Eurycoma longifolia; 3 pChEMBL records | 2026-10-01 |
| D-Limonene | CHEMBL449062 | Plasmodium (screen) | 4.18 | 1 pChEMBL record; ChEMBL is not useful for limonene target affinity | 2026-10-01 |
Not indexed or peptide/biologic¶
| Compound | ChEMBL status | Notes |
|---|---|---|
| BPC-157 | CHEMBL4297358 registered; 0 bioactivity records | Synthetic pentadecapeptide; no curated assay data |
| Apelin-13 ★ | Not indexed — peptide | New wiki page (2026-07-23); stub CHEMBL4650373 exists with 0 activities |
| Thymulin | Not indexed — peptide | Zinc-bound nonapeptide; NONATHYMULIN analogue (CHEMBL2106455) has 0 activities |
| KPV | Not indexed — peptide | Free Lys-Pro-Val tripeptide absent from ChEMBL |
| Resolvin D1 | Not in ChEMBL | SPM not curated as small molecule |
| Maresin 1 | Not in ChEMBL | SPM not curated as small molecule |
| Lactoferrin | Protein/biologic | No ChEMBL small-molecule entry |
| Houttuynia cordata polysaccharides | Complex botanical | No ChEMBL entry |
★ = new to this quarter's sweep.
Discrepancy blocks¶
Salidroside — MAO-B as top ChEMBL target¶
Salidroside (CHEMBL465208, tyrosol-glucose glycoside) was added to the Open Enzyme candidate screen after the 2026-Q3 refresh. The ChEMBL bioactivity set (211 records; 8 with pChEMBL values) shows monoamine oxidase B (MAO-B) as the top annotated protein target: IC50 pChEMBL 6.09 (Ki 6.04), both from CHEMBL4706619. The wiki entry for salidroside (nlrp3-inhibitor-screen.md) cites one MSU animal-model retrieval lead (PMID 30265377) and notes that the primary study is not rehydrated. MAO-B inhibition at ~800 nM is not mentioned in the current wiki. This is not a contradiction of the MSU animal claim, but it surfaces an additional target profile to verify before attributing any salidroside effect solely to NLRP3 or urate-pathway activity. Queue file: synthesis/queue/2026-10-01-chembl-discrepancy-1-salidroside-mao-b.md. (source: ChEMBL refresh 2026-10-01) (In Vitro)
Theaflavin-3,3'-digallate — OATP1B1/OATP1B3 hepatic transporter inhibition¶
The ChEMBL bioactivity set for theaflavin-3,3'-digallate (TF3, CHEMBL434864) contains four pChEMBL-bearing records, all against hepatic uptake transporters from a single study (CHEMBL3039007): OATP1B1 Ki pChEMBL 5.82 (~1.5 μM), OATP1B1 IC50 pChEMBL 5.56, OATP1B3 Ki pChEMBL 5.47 (~3.4 μM), OATP1B3 IC50 pChEMBL 5.35. The wiki page for theaflavins (theaflavins.md) covers renal URAT1/GLUT9/OAT1 expression effects and intestinal ABCG2 substrate/induction data, but does not mention hepatic OATP1B1 or OATP1B3. OATP1B1/1B3 are hepatic sinusoidal importers relevant to drug-drug interactions (statins, methotrexate, urate-lowering agents). Inhibition at these concentrations could affect co-administered drug pharmacokinetics and may be a relevant safety variable in any experimental design using concentrated TF3. Queue file: synthesis/queue/2026-10-01-chembl-discrepancy-2-theaflavin-digallate-oatp1b.md. (source: ChEMBL refresh 2026-10-01) (In Vitro)
What ChEMBL can do here¶
ChEMBL is a discovery and cross-check surface for curated activity records. It is useful for locating a candidate compound–target assay and for noticing that a familiar compound may have a relevant off-target or adjacent mechanism.
It is not, by itself:
- a complete biological-interactor catalogue;
- a transporter-substrate authority;
- a natural-product or multilingual literature census;
- evidence that a non-retrieved relationship does not exist;
- or a basis for ranking values from different assays.
The current Open Enzyme receipt retains version/date context and a refresh recipe, but not immutable raw responses and every exact request parameter from the legacy runs. It therefore cannot support record counts, coverage rates, zero-entry claims, “top target” rankings, or exhaustive absence conclusions.
Match the source to the question¶
| Question | Appropriate evidence source |
|---|---|
| Does compound X bind or inhibit target Y in a named assay? | ChEMBL as a locator, then the primary assay paper |
| Is X a transporter substrate or inhibitor? | Primary transport study, product label, or a curated transporter source with relationship type and cited evidence |
| What physiological reaction or pathway contains X or Y? | Primary biology plus Reactome/KEGG as pathway infrastructure |
| Does a biologic or peptide alter the pathway? | Material-specific biochemical, cell, animal, and clinical literature |
| Does a natural product have relevant evidence? | ChEMBL plus primary multilingual searches using mechanism, material/species, traditional formula, and pathology framing |
Per-record verification¶
Before a ChEMBL-derived claim enters a reader-facing page:
- Save the exact query, database version, access date, filters, and returned record identifier.
- Open the named primary paper.
- Verify compound identity, target, species, assay format, cell system, stimulus, time point, units, qualifiers, and whether the value is measured or inferred.
- State the evidence level next to the claim.
- Keep biochemical, cellular, functional, and phenotypic results separate.
- Do not call a functional pathway readout direct binding.
- Treat a missing record as an unresolved query result.
NLRP3 naming rule¶
- Direct NLRP3 inhibitor: a source-verified direct NLRP3 binding or inhibition measurement in a named assay.
- NLRP3 pathway modulator: a functional inflammasome output or an upstream/downstream mechanism.
This distinction is about what was measured, not whether the lead is interesting. A functional MSU result may be more decision-relevant than a biochemical binding value, but it should keep its actual label.
Cross-assay rule¶
Do not compare or divide potency values when species, cell type, stimulus, endpoint, time, or assay format differ. In particular, separate mouse and human dapansutrile records motivate a matched species-bridging experiment; their legacy numerical ratio is not an isolated species effect and does not explain clinical dosing or efficacy.
Refresh receipt¶
A future refresh should write a compact machine-readable receipt under logs/ containing:
- ChEMBL release and access date;
- exact request URLs or parameters;
- compound and target identifiers;
- filters and pagination;
- returned record identifiers;
- failures and retry state;
- and hashes or retained raw responses sufficient to reproduce any count or non-retrieval statement.
Scientific interpretations belong on the mechanism-owning wiki page after primary-source verification. The receipt records method, not a second findings narrative.
Useful leads that survive¶
Quarterly sweeps have surfaced: quercetin–5-LOX (PMID 2066989, rehydration required), beta-caryophyllene–CB2 (Ki 6.82, CHEMBL4411259), and direct NLRP3 binding records for dapansutrile (IC50 9.00) and oridonin (IC50 6.11). The 2026-Q4 sweep adds two new leads: salidroside–MAO-B (IC50 6.09, CHEMBL4706619) and theaflavin-3,3'-digallate–OATP1B1/OATP1B3 hepatic transporter inhibition (Ki 5.82/5.47, CHEMBL3039007). All are leads to rehydrate from their primary papers. This page does not preserve an assay-wide rank, complete compound table, or claim that these are the only relevant records.
Decision rule¶
A ChEMBL record advances a scientific claim only when its primary assay is verified and the measurement answers the question being asked. A database surprise can create a Research Conjecture; it cannot establish gout relevance, exposure, additivity, safety, or production priority by itself.
Methodology page. Git retains the retired legacy table and its query-era annotations.