🧬 Aging Wiki

finding-compound-data

Properties1
aliasescompound databases, drug data, PubChem, DrugBank, ChEMBL

4 min read

SOP: finding small-molecule and drug data

For any compound page (drug, supplement, metabolite, research chemical), pull from these primary sources.

Identifier resolution (always do first)

Resolve the compound to a canonical identifier before doing anything else.

  1. PubChem CID (the most universal identifier) — search by name at https://pubchem.ncbi.nlm.nih.gov/.
  2. InChIKey — also unambiguous; useful for cross-database lookup.
  3. CAS number — older identifier; some sources still use it.
  4. DrugBank ID (if pharmaceutical) — https://go.drugbank.com/.
  5. ChEMBL ID (for bioactivity data) — https://www.ebi.ac.uk/chembl/.

Synonyms are unreliable (e.g., “vitamin E” refers to a family of 8 compounds). Always pin to PubChem CID first.

Primary databases

PubChem — https://pubchem.ncbi.nlm.nih.gov/

Best for: Structure, physicochemical properties, biological assays, cross-references to other databases. Free, comprehensive.

What to extract:

  • PubChem CID (e.g., 5281614 for fisetin)
  • Canonical SMILES, InChIKey
  • Molecular formula and weight
  • Solubility, logP (relevant to bioavailability)
  • BioAssay summary (high-throughput screening data)

DrugBank — https://go.drugbank.com/

Best for: Pharmaceutical drugs — mechanism of action, indications, pharmacokinetics (absorption, distribution, metabolism, half-life), drug interactions, FDA approval status.

What to extract:

  • DrugBank ID
  • Mechanism of action (1–2 sentences)
  • Targets (linked to UniProt accessions)
  • Pharmacokinetics: bioavailability, half-life, route(s) of administration, hepatic vs renal clearance
  • Approved indications
  • Black-box warnings or major adverse effects

Note: DrugBank is open-access for academic use but requires an account for some data fields.

ChEMBL — https://www.ebi.ac.uk/chembl/

Best for: Bioactivity data — binding affinities (Ki, Kd, IC50, EC50) against measured targets, with literature references. Critical for distinguishing on-target vs off-target effects.

What to extract:

  • ChEMBL ID
  • Top 10–20 measured targets with potency values
  • Distinguish biochemical assays (purified protein) from cell-based assays (more physiologically relevant)
  • Reference paper PMIDs for each affinity value → create studies/ pages for the most-cited

HMDB — https://hmdb.ca/ (for endogenous metabolites)

Best for: Endogenous human metabolites — biofluid concentrations (blood, urine, CSF), normal vs disease ranges, biological roles.

What to extract: HMDB ID; normal blood/serum range; metabolic pathway membership.

Pharmacology-specific

ClinicalTrials.gov — https://clinicaltrials.gov/

Best for: Tracking which compounds are in human trials and at what stage.

What to extract: NCT numbers, phase, status (recruiting / active / completed / terminated), primary outcomes, sponsor.

For aging-relevant compounds, this is often the most up-to-date source on human evidence.

TWOSIDES / OFFSIDES (Tatonetti lab) — adverse effect databases

Best for: Real-world adverse-effect frequencies from FDA AERS data, including drug-drug interactions.

Workflow for a new compound page

  1. PubChem CID + InChIKey + SMILES.
  2. DrugBank (if pharmaceutical) — mechanism, PK, indications.
  3. ChEMBL — actual measured target affinities (don’t trust mechanism-of-action prose without affinity numbers).
  4. ClinicalTrials.gov — current trial status (use NCT numbers).
  5. PubMed/archive search — find the 3–5 most-cited papers on the compound’s aging-relevant effects → create study pages.
  6. Map to Hallmarks of Aging via mechanism.

Frontmatter example

---
type: compound
aliases: [3,7,3',4'-tetrahydroxyflavone, "5,7,3',4'-tetrahydroxyflavone"]
pubchem-cid: 5281614
inchikey: XHEFDIBZLJXQHF-UHFFFAOYSA-N
chembl-id: CHEMBL31096
drugbank-id:
cas-number: 528-48-3
mechanisms: [senolytic, antioxidant, sirtuin-modulator]
measured-targets: ["[[bcl-xl]]", "[[bcl-2]]", "[[pi3k]]"]
hallmarks: ["[[cellular-senescence]]"]
clinical-stage: phase-2-trials
human-evidence-level: limited
half-life: "uncertain (poor oral bioavailability)"
---

Key quality checks for compound pages

  • Bioavailability matters more than potency. A compound with nM potency in vitro but <5% oral bioavailability may produce zero in vivo effect at supplementable doses. Always check PK.
  • Cross-reference dose-response across studies. A senolytic effect at 100 mg/kg in mice does not extrapolate to a 100 mg human supplement. Use allometric scaling and check ChEMBL for the actual exposure-response relationship.
  • Distinguish “supplement” from “drug.” Supplement compounds rarely have rigorous PK data; flag this with #gap/dose-response-unclear.
  • Watch for “polypharmacology” claims. Many natural products hit dozens of targets; an MoA claim limited to one target is usually incomplete.

What NOT to trust

  • Supplement vendor websites for mechanism claims.
  • “Mechanism of action” prose that doesn’t cite measured affinities.
  • In silico target predictions without experimental validation.
  • News coverage of preclinical “miracle compound” findings.

See also

Graph View

Backlinks