PharmacotherapyDB 1.0: Catalog of Drug Therapies for 97 Diseases

PharmacotherapyDB 1.0 is an open catalog of drug therapies for disease, containing physician-curated medical indications. This release contains 97 diseases, 601 drugs, and 1,388 categorized drug-disease pairs, including 755 disease-modifying, 390 symptomatic, and 243 non-indication pairs. The catalog was created by Daniel Himmelstein and integrates DrugBank and Disease Ontology identifiers for analysis.

Use Cases

• Training classification models to predict therapy types (disease-modifying, symptomatic, non-indication) based on drug-disease pairs.
• Integrating drug and disease knowledge graphs using the provided DrugBank and Disease Ontology identifiers.
• Analyzing curator consensus and source agreement for indications using the `n_curators` and `n_resources` fields.
• Benchmarking drug indication extraction algorithms against a physician-curated ground truth.

Strengths

• Contains 1,388 categorized drug-disease pairs derived from physician curation.
• Integrates standardized identifiers from DrugBank and Disease Ontology for 601 drugs and 97 diseases.
• Provides transparency into curation with fields for the number of agreeing curators (`n_curators`) and source resources (`n_resources`).

Limitations

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

Provenance

Source

Daniel Himmelstein via paperswithcode

Collection Method

Physician-curated medical indications compiled from resources including MEDI-HPS, LabeledIn, EHRLink, and PREDICT.