SMILES notation
A compact text format for encoding molecular structures as strings — widely used to represent chemical compounds in AI training data and tool inputs.
SMILES (Simplified Molecular Input Line Entry System) is a standard for writing chemical structures as a single line of ASCII text. Each atom is represented by its element symbol, and bonds, branches, rings, and stereochemistry are encoded with punctuation. For example, ethanol is CCO, caffeine is Cn1c(=O)c2c(ncn2C)n(c1=O)C.
SMILES strings are the most common input/output format for AI chemistry tools because they’re compact, human-readable enough to inspect quickly, and machine-parseable without specialized software. Most molecular databases (PubChem, ChEMBL, ZINC) distribute data in SMILES format, and most AI chemistry models accept or output SMILES.
Why it matters for AI chemistry:
Machine learning models can’t work directly with structural diagrams. SMILES provides a standardized text encoding that lets models be trained on large reaction databases — the same way language models are trained on text. Tools like IBM RXN for Chemistry and REINVENT 4 take SMILES as input for retrosynthesis planning and molecule generation.
Practical limitations:
- The same molecule can be written as multiple valid SMILES strings (canonical SMILES tools like RDKit standardize these)
- SMILES encodes connectivity and some stereochemistry, but not 3D conformation — for 3D structure prediction you need additional tools
- Very large molecules (proteins, polymers) are awkward in SMILES; other formats like FASTA or InChI are more appropriate
Related terms: Molecular Fingerprint, Retrosynthesis
Related guide: Chemistry