Protein Residue Classification using Graph Convolutional Networks

Recent global progress in artificial intelligence has provided scientists with new tools to reliably predict the molecular 3D structures of proteins. In July 2021, DeepMind and the EMBL-EBI collaborated to release the world’s largest and most complete database of predicted protein structures. However, 3D structure alone is not always enough to build a biological understanding and apply these structures to pharmaceutical or functional research programs. This collaborative project between Cyclica and Bo Wang from the University of Toronto and the Vector Institute will build a new AI framework to systematically generate functional predictions for every link (residue) in the protein chain. One such application will be in the detection of ‘druggable’ regions of the predicted protein structure, for use in pharmaceutical drug design applications.

Intern: 
Nasim Abdollahi
Faculty Supervisor: 
Bo Wang
Province: 
Ontario
Partner: 
Partner University: 
Discipline: