Pairwise Molecular Similarities for DrugBank Compounds Using Morgan Fingerprints

Molecular similarities were calculated between pairs of DrugBank compounds using structures from the DrugBank SDF download. The dataset was created by Daniel Himmelstein and is available via the paperswithcode platform. The similarity scores are based on Morgan fingerprints and Dice coefficients.

Use Cases

• Virtual screening for drug repurposing based on structural similarity scores.
• Benchmarking new molecular fingerprinting algorithms against the established Morgan/Dice method.
• Analyzing the chemical space of approved drugs and investigational compounds in DrugBank.

Strengths

• Derived from the authoritative DrugBank database, a key resource for pharmaceutical compounds.
• Uses a well-defined and reproducible cheminformatics method (Morgan fingerprints with radius 2 and Dice coefficients).

Limitations

• Row count and dataset size are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

DrugBank SDF download, processed by Daniel Himmelstein.

Collection Method

Morgan (circular) fingerprints with radius 2 were calculated for each DrugBank compound, followed by pairwise Dice coefficient calculation.