Chem. Id Resolver

The Chemical Identifier Resolver (CIR), commonly reached at its cactus.nci.nih.gov URL, is a public web service operated by the CADD (Computer-Aided Drug Design) Group at the U.S. National Cancer Institute. Its purpose is narrow but widely useful: given one chemical structure identifier, it resolves and converts it into other identifiers or representations. It exposes both an HTML form and a stateless REST-style URL API of the shape /chemical/structure/{identifier}/{representation}, which makes it easy to script from any language or embed in larger pipelines.

CIR accepts a broad range of inputs. It first infers the identifier type, then applies "resolvers" in precedence order until one produces a hit: the CACTVS toolkit decodes line notations such as SMILES, InChI and CACTVS-native strings; a Standard InChIKey lookup queries roughly 100 million indexed keys; the NCI/CADD hashed-identifier database (FICTS/FICuS/uuuuu forms) covers about 96 million unique structures; and a name resolver matches around 68 million names, chaining to OPSIN for systematic IUPAC names before falling back to CIR's local index. It also resolves CAS numbers, NSC numbers, PubChem SIDs, ChemSpider IDs and similar registry codes.

Outputs include standardized SMILES, Standard InChI and InChIKey, IUPAC name and synonym lists, molecular formula and weight, computed properties, 2D depictions (GIF/PNG), 3D coordinates, and SDF and molfile records. Any request can return plain text or, by appending /xml, a structured XML result. Internally CACTVS normalizes structures and generates canonical "unique SMILES," which is what gives CIR value as a canonicalization and cross-referencing hub.

Within the graph-data and format-conversion world CIR is a data source and normalizer rather than a graph tool: it feeds molecular-graph pipelines by supplying clean structures. It pairs naturally with local toolkits such as Open Babel, RDKit and Chem. Structure Conv. for offline conversion, and with clients like the R webchem package that wrap its API. Honest limitations: it is an online service with no uptime or rate guarantees and has seen extended outages; name lookups use exact string matching and may silently drop stereochemistry; the heavier FICTS/FICuS/uuuuu calculations can time out and fall back to raw HASHISY codes; and InChIKey lookups require full three-layer keys. Certain characters (e.g. "#" as %23) must be URL-escaped.