Drug Repositioning based on machine learning

Here you can find new potential indications for a given drug or small molecule compound.

The prediction is based on machine learning models, which were created on the basis of molecule structures in Promiscuous 2.0 (mapped to indications using the "Therapeutic Target Database"). Querying a structure of interest will be performed by evaluating it against all ICD-10 categories, for which there were enough structures to build a well-performing model, and displaying all categories which are predicted as potentially associated.

Since a similarity search is performed over the drugs in the database, a molecule structure is required as input. There are different ways to specify your structure of interest:

1a. Search a compound by its name via PubChem (the corresponding structure will be uploaded after clicking the start button)
1b. Create a structure by a SMILES string
1c. Draw your structure with the drawing tools
1d. Load a molecule by clicking on

2. Once a structure was loaded into the ChemDoodle Viewer, you may (but not must) make changes to it using the drawing tools.
3. When you are satisfied with the structure, start the similarity search by clicking the button below the structure viewer.

Name: e.g. Warfarin
e.g. CC(=O)CC(C1=CC=CC=C1)C1=C(O)C2=C(OC1=O)C=CC=C2

Please note that the calculation might take up to 1 minute!

The content of this website is intended for research purposes only and in no way to be interpreted as professional medical advice.

Successful Drug-Repositioning

Sildenafil Thalidomide

About Overview

Following you can find some statistics and further information about this site

Contact Us

In case you have some questions, please feel free to contact us personally.

Contact: Site