Over the last years the world of drug discovery has changed significantly as large amounts of high quality data have appeared in the public domain. Big databases appeared, containing chemical structures (compounds), the (protein) targets they interact with, and information of the strength of the interaction (affinity).  Computer models are used to understand why certain compounds bind to certain proteins. These models can then be used to predict activity of new (unknown) compounds on protein targets. This can be used to virtually screen for new potential drugs.

In the Medicinal Chemistry group, a collection of compounds have been synthesized over the last few decades. These compounds have been tested against a number of proteins in the projects wherein they were synthesized but beyond that, knowledge on their biological activity is limited.

This project consist of two parts. Firstly the existing compound collection will be standardized and added to a local copy of ChEMBL (for ease of use). Secondly an ensemble of models that were trained on the ChEMBL database (version 27) will be used to predict their biological activity spectrum. 

Share on Facebook Share on X Share on LinkedIn Share by WhatsApp Share by Mastodon