Accelerated drug discovery

Investigating how new approaches to Artificial Intelligence (AI), such as geometric deep learning and graph neural networks, can help accelerate the development of new drugs for neurodegenerative diseases, such as ALS, multiple sclerosis, Parkinson’s disease or Alzheimer’s disease.

project
Description of the project

There are currently no reliable and accurate computational techniques to predict the outcome of kinase panels performed in vitro, which are common in the traditional processes of discovering proteins with pharmacological activity. Although considerable effort has been made in the last two decades in research and development of tools to support biologists in the discovery of new drugs, the problem of in silico estimation of affinity measurements used in the laboratory remains largely unexplored today. Our research branch will use novel techniques in the field of Graph Neural Networks (GNN), which will allow to efficiently utilize not only the chemical and physical (including quantum) properties of both the molecule and the protein, but also the three-dimensional structure of both.

The technology resulting from this research orbital will make it possible to replicate on computer the experimental affinity tests between potential drugs and protein kinases. In other words, it will allow most laboratory affinity tests to be replaced by in silico tests. The result will be a drastic reduction in the time and cost of molecule-kinase affinity testing and, ultimately, an increase by several orders of magnitude in the number of compounds that can be considered in a single study. Such a technique will also open the door to new, more optimized, and automated processes in which new drug researchers, with the help of this computational technique, will perform massive searches for both known and new small molecules using known and useful metrics.

tech
IA
Biotechnology
collaborating
Collaborating entities
University of Columbia
CICbioGune