x

Repurpose Drug catalogue

RepurposeDrugs

RepurposeDrugs is another venture of BioICAWtech in collaboration with scientists from University of Helsinki.

 

RepurposeDrugs is a web-portal that combines a unique drug indication database with a machine learning (ML) predictor to discover new drug-indication associations for mono and combination therapies.

Request a demo
Figure: The RepurposeDrugs landing page snapshot, displaying an interactive heatmap of drug-disease associations with diseases color-coded by high-level disease groups (left panel) and a word cloud plot highlighting drugs with the most approved disease indications (bottom right side)
Figure: Comparative distribution of (a) unique drugs (or compounds) and (b) disease indications in different developmental stages: approved, terminated and under investigation in clinical phases I, II, and III. (c) Number of approved drugs common across various disease groups. (d) Number of approved drugs, with failed clinical trials, common across different disease groups.

It allows users to examine treatment approval statuses, indications, and details on ongoing or terminated clinical trials across 25 disease categories, including neoplasms, skin, and cardiovascular conditions.

 

The portal catalogs 4,314 drugs, covering approved, terminated, and investigational compounds, and features 161 drug combinations, associated with 1,756 diseases/conditions, resulting in 28,148 drug-disease indications. Accessible without any login requirements, RepurposeDrugs offers a user-friendly interactive exploration of drug-disease associations, allowing data export in PNG and SVG formats.

Figure: RepurposeDrugs prediction workflow consists of three main steps. Left panel: Manual curation of datasets, integrating drug/disease identifiers, primary targets, structural information, and disease categorization. Middle panel: Model training and testing, highlighting feature extraction from drug and disease descriptors. Right panel: New drug-disease associations predictions, employing a conformal prediction approach to exclude low-confidence predictions.

Beyond its extensive data repository, RepurposeDrugs also offers Machine Learning-powered predictions for potential drug-disease associations for both single and combination treatments. The underlying algorithm is trained on known approved/failed indications and capable of predicting disease association scores for a broad spectrum of user-provided compounds, regardless of their development stage. Its performance, using out-of-fold predictions, has demonstrated Pearson (point-biserial) correlation of 0.75 and 0.56 for mono and combinations respectively. Soon, we will integrate more datasets in RepurposeDrugs database and further improve the prediction algorithms.

 

To our knowledge, RepurposeDrugs is the first integrative database for both single-agent and combination therapies, featuring an ML-based web tool for predicting the likelihood of treatment approval. Unlike conventional drug repurposing databases, it covers a broad range of diseases and a variety of both single drugs and combinations, opening new avenues for researchers and clinicians seeking innovative therapeutic solutions.