PubChem PUG REST API - Chemical Structures & Properties

The reference shelf for small-molecule chemistry, filed as one addressed dataset inside Datadory's life sciences tools & services slice. Behind it sits NCBI's PubChem: roughly 115 million compounds, 300 million substances and 900 million bioactivity data points, organized on a clean three-key entity model - CID for compounds, SID for substances, AID for assays - that also reaches gene, protein, pathway, taxonomy and cell entities.

Thirty-five verified fields define the compound row: identity and nomenclature (formula, three identifier systems, generated IUPAC name), computed descriptors (XLogP, exact and monoisotopic mass, topological polar surface area, complexity, charge, hydrogen-bond and rotatable-bond counts), stereochemistry tallies, 3D conformer geometry, a substructure fingerprint, plus synonym, patent-count and literature-count layers. Eight response shapes ride along, JSON through CSV to SDF structure files and PNG-rendered depictions.

It scores 10 out of 10 on Datadory's rubric against a catalog averaging 7.81 across 1,744 datasets - a mark only 145 records earn, and it plays the compound half beside UniProt REST API's protein half. Get a sample of this dataset cut to your own compound list.

One standardized compound per row, identity pinned first and computed values riding behind. Three everyday molecules, straight from the cataloging capture:

# delivered grain: one compound record per row

CID             : 2244
MolecularFormula : C9H8O4
MolecularWeight  : 180.16
IUPACName        : 2-acetyloxybenzoic acid

CID             : 2519
MolecularFormula : C8H10N4O2
MolecularWeight  : 194.19
IUPACName        : 1,3,7-trimethylpurine-2,6-dione

CID             : 1983
MolecularFormula : C8H9NO2
MolecularWeight  : 151.16
IUPACName        : N-(4-hydroxyphenyl)acetamide

Read the top row as three decisions already made upstream. Aspirin arrives as CID 2244 - the stable numeric address every other table joins on - with its formula already normalized to Hill notation and its weight carried to two decimals, not parsed out of prose. The caffeine row proves the dictionary holds heteroatom-rich ring systems as cleanly as substituted benzoic acids, and the paracetamol row shows the generated IUPAC name doing its job where marketing names diverge.

A fourth captured layer shows the annotation side. The same CID 2244 carries Title Aspirin alongside a description line attributed to its contributing source - the California Office of Environmental Health Hazard Assessment - so quoted text travels with provenance attached instead of anonymized. Synonym lists behave the same way: trade names, CAS numbers and systematic names arrive as an array, ready to match against whatever inventory list you hold.

Every value above comes from the catalog's August 2026 research pass, not a mock-up.

Thirty-five documented fields define the compound row, each definition verified during the August 2026 research pass - a bar only 145 of 1,744 cataloged datasets clear outright. Around fifty property tags exist in total; the remaining descriptor tags fold into your sample on request alongside derived columns such as Lipinski-style rule-of-five flags built from the counts below. They split into six jobs:

• Identity: CID anchors every row, joined by Title, MolecularFormula and MolecularFormulaNoCharge.
• Structure representations: SMILES with stereo and isotope layers, ConnectivitySMILES without them, standard InChI, the 27-character InChIKey hash and the base64-encoded Fingerprint2D substructure fingerprint.
• Nomenclature: generated IUPACName plus the Synonym array - trade names, CAS numbers, systematic names in one field.
• Mass and lipophilicity: MolecularWeight, ExactMass, MonoisotopicMass and computationally generated XLogP.
• Drug-likeness descriptors: TPSA polar surface area, Complexity, formal Charge, HBondDonorCount, HBondAcceptorCount, RotatableBondCount and HeavyAtomCount.
• Stereo, 3D shape and linkage: atom- and bond-level stereo counts, defined versus undefined splits, CovalentUnitCount, Volume3D, FeatureCount3D, ConformerCount3D, ConformerModelRMSD3D, then PatentCount, PatentFamilyCount, LiteratureCount and source-attributed Description text closing the record.

The full dictionary follows in tabular form below.

• Geography - global by construction: chemical deposits and consolidated literature links arrive from sources worldwide, and no compound carries a jurisdiction flag. Region-shaped questions resolve by joining the patent and literature counts outward, not by filtering rows directly.
• Temporal - compounds reach back through historical deposits, with assay deposition running continuously since PubChem's 2004 launch and the corpus refreshed continuously since. Records are current-state: today's identifiers, today's calculated values, today's link counts.
• Granularity - four levels deep: whole compounds keyed by CID, deposited substances keyed by SID beneath them, individual assay results keyed by AID, and gene, protein, pathway, taxonomy and cell entities hanging off the same spine. Whole-corpus cuts (every compound under 500 Da with fewer than five hydrogen-bond donors) and narrow ones (one molecule's full synonym and assay neighborhood) both resolve without touching non-matching rows.

Set against the wider catalog - 1,744 datasets averaging 7.81 - the 10/10 prices in verified definitions, three clean join keys and a 35-field dictionary that stays flat at scale. Where it ranks: best life sciences tools services datasets.

API, files, or your warehouse. Daily, weekly, or hourly.

You pick the channel and cadence; normalizing the identifier graph, keeping standardized compounds apart from raw deposits and flattening fifty-plus descriptor tags into joinable columns stay our problem. Rows arrive CID-keyed, so this month's delivery appends cleanly onto last month's and joins whatever assay, target or inventory tables you already hold.

Hourly suits screening pipelines triaging newly deposited candidates while they are still decisions rather than history. Daily fits competitive monitoring of patent and literature link movement across a compound set. Weekly matches the rhythm most cheminformatics workflows actually run at. Whichever you choose, the thirty-five-field dictionary travels unchanged. A sample ships first either way - real rows for your own molecule list before any commitment.

Six jobs the compound shelf settles outright:

1. Candidate triage before synthesis - filter 115 million compounds on XLogP, TPSA, hydrogen-bond donors and rotatable bonds to rank-order molecules worth ordering before anyone books instrument time; see ml model training for the model-building variant.
2. Structure lookup and identifier resolution - turn names, CAS numbers or registry synonyms into canonical CIDs, SMILES and InChIKeys, or run the reverse; see api integration.
3. Inventory reconciliation - match an internal catalog against canonical structures on InChIKey, surfacing duplicates, stale registrations and mislabelled stock without name-based guesswork.
4. Virtual screening and similarity work - the substructure fingerprint plus stereo-aware identifier layers make neighbor searches a computed column rather than a chemistry engine you have to stand up.
5. Market and landscape sizing - compound, substance and assay counts frame cheminformatics and compound-management market sizings; the same job serves the wider market sizing shelf.
6. Citation-grade verification - quote formula, weight and generated names straight from the reference record when a paper or filing needs a number that holds up; see citation grade research.

Each job maps to a persona below, and the sample validates whichever one you came for.

Data Scientists & ML Engineers lead fit at relevance 3: decades of standardized structures with computed descriptors attached is the combination molecular-property models rarely get in one place - see life sciences tools services data for data scientists. Developers & Data-Product Builders also tag relevance 3, wiring stable CIDs into chemistry lookups and lab software where a number survives redesigns far better than a name ever will - see life sciences tools services data for developers builders. Journalists, Academics & Students (relevance 3) cite canonical structures and generated names with provenance attached. Competitive Intelligence & Product Teams and Market Researchers & Consultants (both relevance 1) read deposition and link counts as activity signals and market denominators.

Persona fit has edges worth naming: this describes structures; it does not measure expression, toxicity outcomes or clinical progress. Those live in neighboring catalogs - measured activities at ChEMBL, transcriptomic evidence at NCBI GEO, trial evidence at ClinicalTrials.gov. Pair accordingly rather than expecting one feed to cover the whole discovery chain.

Three honest edges. First, standardized versus deposited: the 115-million compound count is deduplicated CIDs, while the roughly 300 million substance records count deposits as submitted - one molecule can carry many SIDs, so pick the grain that answers your question rather than the bigger number. Second, calculated means calculated: XLogP, TPSA and the descriptor family are model-generated values, not measurements, and any workflow needing measured biology should pair this record with a curated-activity source. Third, current-state only: trendlines over link counts or descriptor revisions come from repeated captures on a fixed schedule - which is what the daily, weekly or hourly cadence exists to automate. None of this is hidden at delivery; say which edge bites your question and the sample comes back shaped around it.

Cards worth opening next:

• Life Sciences Tools & Services data hub - the pooled industry view this record files under, three primaries deep.
• Best life sciences tools services datasets - where this perfect score ranks inside its industry.
• NCBI PubChem source profile - steward background and the deliverables cut from this corpus.
• UniProt REST API - Protein Sequence & Function - the protein half of the slice, quality 10 beside quality 10.
• PubChem PUG REST API vs UniProt REST API - compound identity against protein annotation, scored honestly.
• SelectScience - Lab Product Reviews & Catalog - the instruments-and-reagents market itself, if your question is vendors rather than molecules.
• ChEMBL and RCSB Protein Data Bank - measured bioactivity and experimental 3D structure for the same molecules and targets.

Then get a sample of this dataset - real rows cut to your own molecule list before any commitment.