Overview
This page documents how the atlas resolves the identity, sequence, and chemical structure of each catalog entry. Many entries sold under a single name are not a single well-defined substance; the goal is to record what authoritative databases say, not what any vendor claims.
Three identity caveats
A database record is a registry pointer, not proof of a commercial sample's identity. A two-dimensional depiction cannot establish three-dimensional or formulation context. When authoritative evidence does not resolve a unique entity, the atlas retains the ambiguity instead of inventing a structure.
Resolution moves from marketed or common name through preferred identity and aliases, sequence and modifications, chemical form, stable identifiers, and asset provenance. Each layer narrows meaning. An undefined mixture or unresolved identity is documented as a boundary, and no structure is invented.
Biological sequence versus chemical structure
Peptides and proteins are linear polymers of amino acids. Their primary structure (sequence) is the order of residues. For small peptides (under 40 residues), PubChem typically records a defined chemical structure: the sequence of atoms and bonds. For larger peptides and proteins, the stored CID corresponds to a simplified atomic representation or a sequence-only record; a 2D structure image cannot capture the three-dimensional fold, disulfide pairing, or post-translational modifications.
The atlas draws the following distinction:
Sequence source: an authoritative source for the amino-acid sequence (DrugBank, UniProt, FDA/EMA label, or PubChem Compound summary).
Structure source: the The record number of a compound in PubChem, the atlas's primary structure-asset source. A registry record or depiction does not authenticate a commercial sample. Source de la définition: Identity and structure assets methodology · Glossaire from which the atlas retrieves SMILES for a locally generated 2D depiction. The depiction is a 2D rendering of the structural formula, not an experimentally determined conformation.
The fuller identity workflow is also explained in How Peptide Identity Is Verified.
Sequence or residue order must be considered with termini, cyclization, crosslinks, stereochemistry, conjugates, and salts or complexes before a context-specific chemical entity can be depicted. TB-500, GHK-Cu, thymalin, and cerebrolysin retain their documented ambiguity.
Authoritative databases consulted
| Source | Identity role | What it can establish | Limitation |
|---|---|---|---|
| PubChem | Primary structure-asset source | CID assignment, record title, and the SMILES used for a locally generated depiction, verified through PUG REST | A registry record or depiction does not authenticate a commercial sample |
| DrugBank | Drug identity source | Sequence and pharmacological identity for approved drugs | It does not transfer one product identity to differently marketed material |
| UniProt | Protein identity source | Protein sequence and family classification | A protein record does not define every modified, conjugated, or marketed form |
| ChEBI | Ontology source | Small-molecule ontology identifiers | An ontology identifier alone does not establish sample composition |
| FDA/EMA product labels | Product-specific official source | Official sequence and composition for approved therapeutics | The statement remains product- and jurisdiction-specific |
| ClinicalTrials.gov | A clinical-stage candidate with scientific or public relevance. Investigation is not approval, and a studied exposure is not a recommendation. Source de la définition: Scope and selection methodology · Glossaire record | Candidate sequence or identity where the record discloses it | Public records may not disclose enough detail to resolve the entity |
For PubChem-derived assets, each retrieved SMILES record and locally drawn 2D SVG has a cached input and checksum provenance sidecar.
Salts, conjugates, complexes, and mixtures
Many catalog entries are not single, neutral peptides:
Salts: peptide hydrochloride and acetate salts, including glatiramer acetate. The PubChem record for the free base is used; salt forms are noted.
Fatty-acid conjugates: semaglutide (C18 diacid), liraglutide (C16 palmitoyl), and palmitoyl cosmetic peptides. The conjugate is integral to the molecule; the The record number of a compound in PubChem, the atlas's primary structure-asset source. A registry record or depiction does not authenticate a commercial sample. Source de la définition: Identity and structure assets methodology · Glossaire covers the full conjugate.
Metal complexes: GHK-Cu, a copper(II) complex. The CID includes the copper ion.
Mixtures: thymalin and cerebrolysin, undefined peptide mixtures from tissue extracts. No single CID or sequence applies; no structure image is retrieved.
Random copolymers: glatiramer acetate. No defined sequence applies, so no structure image is retrieved.
Stereochemistry
PubChem 2D depictions can encode assigned stereochemistry through wedge/hash
bonds and related annotations. They are nevertheless flattened depictions, not
three-dimensional conformations, and they do not guarantee that every relevant
stereocentre, atropisomer, salt form, or product-specific configuration has
been resolved. Many entries contain D-amino acids, including D-Phe, D-Trp, and
D-Arg in octreotide, ipamorelin, and cetrorelix, that are critical for activity.
Identity review therefore uses the registry record and the documented
modifications_conjugates field, not the PNG alone.
Limitations of 2D depictions
2D images flatten three-dimensional structure entirely.
Disulfide bridges, including those in oxytocin, ziconotide, and linaclotide, are shown as S—S bonds but the image does not indicate spatial arrangement.
Cyclic peptides, including cyclosporine, daptomycin, and bacitracin, are depicted as planar cycles, not as complex macrocyclic folds.
Glycosylation in vancomycin and dalbavancin is shown as attached sugars; assigned stereochemistry may be drawn, but a flat image does not establish product composition, conformation, or analytical identity.
Retrieval from PubChem uses the PUG REST property endpoint. RDKit draws the returned SMILES locally. This is a depiction, not an analytically determined structure.
For proteins and large peptides without a stable defined small-molecule structure, including dulaglutide, mecasermin, and follistatin-344, no 2D depiction is retrieved. These entries are marked
not_applicableinassets/structures/manifest.csv.
Registry identity and sample authenticity are different questions. The product quality, authenticity, and testing guide explains why specialized analytical methods and validated specifications are needed to assess material rather than a database record.
Boundary cases explicitly unresolved
| Entry | Ambiguity | Asset decision | Evidence consequence |
|---|---|---|---|
| Modified GRF (1-29) and CJC-1295 | Sequence variants of GHRH; the exact marketed product may differ | Retain the variant boundary rather than using one depiction as generic | Evidence for one defined variant is not silently transferred to another |
| TB-500 | The marketed name may mean full-length thymosin beta-4 or a fragment | Keep the TB-500 identity and asset boundary explicit | Claims remain tied to the studied material, not the ambiguous market name |
| IGF-1 LR3 and Des(1-3) IGF-1 | Modified protein sequences are not captured by a single CID | Do not assign a single CID-derived depiction as definitive | Registry evidence cannot resolve the marketed entity by itself |
| PEG-MGF | PEGylated protein without a defined small-molecule structure | No defined small-molecule depiction | Structural and evidence claims retain the unresolved conjugate boundary |
| GHK-Cu | CID 133697840 represents a bis(GHK)-copper 2:1 species, not the generic 1:1 complex often intended by the name | Retain the CID as a registry pointer but do not use its depiction as the generic entry asset | Evidence is not generalized across differently defined complexes |
| Apraglutide and cagrilintide | A clinical-stage candidate with scientific or public relevance. Investigation is not approval, and a studied exposure is not a recommendation. Source de la définition: Scope and selection methodology · Glossaire sequences are not yet fully disclosed in public regulatory filings | Record the public-identity limit rather than filling the gap | Claims cannot rely on an invented sequence or structure |
Asset management
Generated 2D and 3D asset classes
| Class | File/provenance requirement | Refresh trigger |
|---|---|---|
| Generated 2D SVG | PubChem SMILES or verified sequence input, source sidecar, checksum, and manifest row | A stable record, sequence, identifier, or depiction input changes |
| Idealised 3D still or turntable | Sequence-build provenance plus the required idealised-conformer label and recorded omissions | The verified sequence, modifications, or recorded approximation changes |
| Blend still | Component provenance, shared-scale record, and unresolved-component notes | A component identity, scale basis, or omission record changes |
RDKit generates the 2D SVG class from a PubChem SMILES record or, where no CID
is recorded, from a verified peptide sequence. PyMOL generates the 3D still and
turntable class by building an idealised sequence conformer; blend stills place
those component models on a shared scale. The 3D class is neither an
experimental structure nor a structure prediction. Every displayed 3D asset
therefore carries the label Idealised conformer built from sequence; not an
experimental or predicted structure. Conjugates, terminal decorations,
noncanonical-residue approximations, unresolved blend components, and other
omissions are recorded in the adjacent provenance sidecar.
Each retrieved image in assets/structures/ has:
a corresponding
<slug>.source.jsonprovenance file recording the database, stable ID, retrieval URL, date, SHA-256, and license terms;an entry in
assets/structures/manifest.csvrecording those fields.
Checksums are verified before commit.
Refresh policy
Identities and structure assets are verified as of 2026-08-06. PubChem records may be updated; sequences from DrugBank and UniProt should be rechecked for major revisions. Mixture-based entries—thymalin, cerebrolysin, and glatiramer— have no CID to refresh.