Chemical Representations

Industry blog covering molecular representations, QSAR best practices, and dataset curation, with Jupyter notebooks for each post.

Talktorials (talk + tutorial) covering cheminformatics from SMILES to fingerprints to ML. Jupyter notebooks with step-by-step explanations.

Official RDKit "Getting Started" guide covering molecule reading, fingerprint generation, substructure search, and descriptor calculation.

Collection of practical code recipes for common cheminformatics tasks: SMILES parsing, fingerprint computation, similarity search, and 2-D depiction.

Notebook-based introduction to molecular featurization in DeepChem: Morgan fingerprints, graph features, and Coulomb matrices.

Original ECFP/Morgan fingerprint paper. Essential reading for anyone using fingerprint-based ML.

Introduces SELFIES — a molecular string notation that guarantees chemical validity by construction.

Comprehensive review of molecular representations and ML approaches for QSAR, covering descriptors, fingerprints, and graph representations.

The standard open-source cheminformatics library. Covers SMILES parsing, Morgan fingerprints, 2-D/3-D descriptors, substructure search, and molecular I/O.

Python library computing 1800+ molecular descriptors. Faster than RDKit for large-scale descriptor generation.

High-level molecular manipulation library wrapping RDKit with a cleaner API. Useful for batch featurization and dataset curation.

NIH's open chemistry database: 115M+ compounds with SMILES, InChI, bioactivity data, and 2-D/3-D structures. The primary public source for molecular data.

EBI's curated database of 2.4M bioactive molecules with standardised SMILES, assay data, and physicochemical properties — widely used for representation benchmarking.

700M+ commercially available compounds in SMILES format with physicochemical filters. Standard source for large-scale molecular representation experiments.