Compound Identifiers

Overview

The table below shows the currently available chemical identifiers in the ORD schema. A CUSTOM option is included for capturing other chemical identifiers which are not yet directly included in the schema. While multiple identifiers can be included for a single chemical, users are actively encouraged to include SMILES or InChI identifiers in their datasets. These line notations depict the structure of the chemical and allow the ORD data browsers to resolve images for the chemicals using RDKit.

Identifier	Description
SMILES	Simplified molecular-input line-entry system.
INCHI	IUPAC International Chemical Identifier.
MOLBLOCK	Molblock from a MDL Molfile V3000.
IUPAC_NAME	Chemical name following IUPAC nomenclature recommendations.
NAME	Any accepted common name, trade name, etc.
CAS_NUMBER	Chemical Abstracts Service Registry Number (with hyphens).
PUBCHEM_CID	PubChem Compound ID number; link.
CHEMSPIDER_ID	ChemSpider ID number; link.
CXSMILES	ChemAxon extended SMILES; link.
INCHI_KEY	IUPAC International Chemical Identifier key
XYZ	XYZ molecule file
UNIPROT_ID	UniProt ID (for enzymes); link.
PDB_ID	Protein data bank ID (for enzymes); link.
AMINO_ACID_SEQUENCE	Amino acid sequence (for enzymes).
HELM	Hierarchical Editing Language for Macromolecules (HELM); link.
MDL	MDL number e.g., MFCD00005972 for morpholine, often included on commercial supplier listings.
CUSTOM	Create your own custom identifiers. Include an informative name in the description field.

When preparing an ORD dataset from a paper, an often overlooked task is the collection of SMILES (or InChI) strings for the chemicals which may have been recorded in the paper using names, abbreviations and/or images. This guide suggests some common workflows for collecting these identifiers.

Important

Special care needs to be taken with the identifiers for transition metal catalysts and other organometallic compounds. There are some challenges associated with generating SMILES or InChI strings for organometallics, and this can cause problems with the way RDKit parses and resolves them. ORD authors should always check that their SMILES or InChI strings are being resolved correctly, and please refer to Dative Bonding in Organometallic Compounds below for a recommended workflow for including dative bonding in SMILES strings.

For compounds which are hard to define using existing identifiers, it can also be helpful to include multiple identifiers (e.g. a NAME, CAS_NUMBER and SMILES) to remove some of the ambiguity.

Getting SMILES or InChI Strings

Draw the Structures

Most chemical drawing packages can export a SMILES or InChI string for chemical structures you have selected. If you have already drawn chemical structures for a published article then it can be a simple task to export SMILE or InChI strings from your ChemDraw (or other chemical drawing tool) files. This ChemSpider Blog shows how to do this in Avogadro, ChemDoodle, ChemDraw, ChemSketch and MarvinSketch.

The following online chemical drawing tools can also be used to obtain SMILES or InChI strings:

• Open Babel webservice (generate SMILES or InChI)

• InChI web demo (generate InChI only)

Important

There are some challenges associated with generating SMILES or InChI strings for transition metal catalysts and other organometallic compounds. Please refer to Dative Bonding in Organometallic Compounds below for specific guidance on using chemical drawing packages to obtain SMILES or InChI for such structures.

Look Them Up

The ORD Reaction Editor has a look up function which can be used to add chemicals by name. The name_resolve function uses the PubChem and OPSIN APIs to look up names and returns a SMILES string when it is available.

For looking up identifiers of organic compounds the following online services can be useful:

• PubChem

• OPSIN

• ChemSpider

Important

Online look up of SMILES is not recommended for transition metal catalysts and other organometallic compounds. PubChem’s “canonical” SMILES for metal complexes treat each substructure as a separate component (e.g., metals, ligands, ions), leading to a salt-like representation of the compounds, and most other websites will also report the SMILES in this style. Please refer to Dative Bonding in Organometallic Compounds below for a recommended workflow for generating SMILES strings which include dative bonding.

Lists of names can be programmatically looked up using the ORD name_resolve function in Python.

Converting Identifiers

For converting between chemical identifiers the following online services can be useful:

• NIH NCI/CDD Chemical Identifier Resolver

• Open Babel webservice

Image Extraction

New consumer tools for optical chemical structure recognition are appearing rapidly and could be a useful method for extracting structural identifiers when the above methods are not possible. We have had some success with using Mathpix for extracting SMILES and InChI strings from images already.

With all these tools please take extra care to check that the identifiers are resolved to the correct chemical structures. See Checking SMILES or InChI Strings below for some recommended workflows for checking chemical identifiers.

Dative Bonding in Organometallic Compounds

There are some challenges associated with the generation of canonical SMILES and InChI strings for organometallic compounds such as transition metal catalysts. While the above methods can be used to obtain a SMILES, extra care needs to be taken to check that the SMILES is parsed and resolved correctly by RDKit.

Some problems to watch out for when obtaining SMILES for organometallic compounds include:

• RDKit SMILES parsing fails for ChemDraw-generated SMILES of plainly drawn metal complexes due to the valence rules on N atoms.

• Using dative bonds in ChemDraw fixes this problem but generates charged species to satisfy the valence rules.

• ChemDraw’s syntax for ferrocene complexes leads to RDKit parsing failures.

• PubChem “canonical” SMILES for metal complexes treat each substructure as a separate component (e.g., metals, ligands, ions), leading to a salt-like representation of the compounds.

• Many online sources of SMILES give the salt-like representation used by PubChem.

• With previous versions of InChI the metal complexes are disconnected and represented as a salt.

The current best practice for generating organometallic SMILES for usage in the Open Reaction Database, is as follows:

1. Use ChemDraw (or other chemical drawing tool) to draw the structure with neutral single bonds to ligands. Ferrocenes should be explicitly drawn as cyclopentadienyl-anion-sandwiched Fe atoms.

2. Copy the structure as SMILES (see ChemSpider Blog) for how to do this in Avogadro, ChemDoodle, ChemDraw, ChemSketch and MarvinSketch.

3. Generate dative ‘<-’ bonds programmatically using the set_dative_bonds message helper.

Here’s a snippet of code to generate the dative bonds for a single SMILES string in a Jupyter/Colab notebook:

In [1]: # Import packages
In [2]: from ord_schema import message_helpers
In [3]: from rdkit import Chem

In [4]: # Enter your SMILES string
In [5]: smiles = 'O=S(C)O[Pd]([NH2]C(=C1C=C2)C=C2)(C(=CC=C2)C1=C2)[P](C(C=C1)=C(C(=C(C=C2)P(C(=CC=C3)C=C3)C(=CC=C3)C=C3)C(=C2C=C2)C=C2)C(=C1C=C1)C=C1)(C(C=CC1)=CC=1)C(=CC=C1)C=C1'

In [6]: # Use RDKit to generate a mol for the SMILES string
In [7]: mol = Chem.MolFromSmiles(smiles, sanitize=False)
In [8]: # Use the ORD message_helper to add dative bonds to the mol
In [9]: dative_mol = message_helpers.set_dative_bonds(mol)
In [10]: # Use RDKit to convert the mol back to a SMILES
In [11]: canonical_smiles = Chem.MolToSmiles(dative_mol)

In [12]: # Print the canonical SMILES string
In [13]: print(canonical_smiles)
Out [13]: 'CS(=O)O[Pd]1(<-P(C2=CC=CC=C2)(C2=CC=CC=C2)C2=C(C3=C(P(C4=CC=CC=C4)C4=CC=CC=C4)C=CC4=C3C=CC=C4)C3=C(C=CC=C3)C=C2)<-NC2=C(C=CC=C2)C2=CC=CC=C21'

For a more detailed example refer to this Jupyter Notebook which shows how to use the above code to canonicalize a single SMILES, or to batch process multiple SMILES strings input as a .csv file. The example also shows how to use RDKit to visualize the generated SMILES so you can check that the dative bonds have been generated correctly.

InChI for Molecular Inorganics

The InChI Trust are actively working on adding support for molecular inorganic compounds to InChI. For a recent paper about this see Faraday Discuss., 2025, 256, 503-519, DOI: 10.1039/D4FD00145A and you can try out the development features in the InChI web demo. Switch the version to “Latest with Molecular Inorganics” to test it out on your organometallic structures.

While the Open Reaction Database are following these developments closely, the current advice is to continue using SMILES with programmatic addition of dative bonding as the primary identifier for organometallics. Once the inorganic support is formally adopted into a stable InChI version, and the RDKit Chem package has been updated to resolve the these InChIs, then we will revisit this guidance.

Checking SMILES or InChI Strings

For checking individual identifiers of organic compounds the following online services can be useful:

• SMARTS Plus (SMILES only)

• NIH NCI/CDD Chemical Identifier Resolver

Individual SMILES and InChI identifiers can also be checked in the ORD Reaction Editor. Create a test dataset and add each identifier as a new chemical input component. The ORD editor uses RDKit to parse the identifier and a preview of the structure is generated.

SMILES and InChI strings can also be visualized programmatically using the RDKit Chem Draw package. See this Jupyter Notebook for example code to visualize SMILES strings input individually, or batch input as a .csv file. For reviewing complex and/or large batches of chemical identifiers as part of an ORD dataset preparation, it is also worth speaking to the ORD support team (email help@open-reaction-database.com) to get bespoke advice on your task.