OpenPredict KP

Back to Home

Description

The OpenPredict API can be queried for a pre-computed set of drug-disease pair predictions and drug-drug similarity scores from graphs embeddings

Data sources

The current model includes embeddings from:

• DrugBank-OMIM Predict reference dataset
• The NeuroDKG, a curated database of neurological diseases and the drugs that can be used to treat them
• Embeddings generated from clinical co-occurrence counts from COHD

New embeddings can be easily added in JSON format via the OpenPredict API

Work in progress:

• Integrate other KPs data
• Curated drug indication dataset from DailyMed.
• A curated database of repositioned drugs used off-label curated from the literature

Key metrics

• Prediction score from machine learning model (range: [0-1])
• Embedding similarity score (range: [0-1])

Technical User Guide

The OpenPredict API is maintained by the Clinical Data Knowledge Provider

Modes of Access

• Translator Reasoner API v1.0.0-beta, registered as a SmartAPI
  • Bespoke OpenAPI at https://openpredict.semanticscience.org

Use Cases

• Notebook examples: https://github.com/MaastrichtU-IDS/translator-openpredict/blob/master/docs/openpredict-examples.ipynb

Knowledge Sources Accessed

• OMIM
• DrugBank
• COHD

Source Code

• OpenPredict API: https://github.com/MaastrichtU-IDS/translator-openpredict

External Documentation

• The testing process is described at https://github.com/MaastrichtU-IDS/translator-openpredict/blob/master/TESTING.md
• Documentation to contribute: https://github.com/MaastrichtU-IDS/translator-openpredict/blob/master/CONTRIBUTING.md

Contact

• Vincent Emonet