University of Ottawa

Published purpose

Motifs are active structural sites tasked with beneficial biological functions that can be employed to create new vaccines and enzymes. When designing proteins, a crucial endeavor is the creation of stable scaffolds to support motifs, thus allowing them to maintain their shapes and, consequently, their function. Unfortunately, current machine learning-based techniques are timeconsuming and limited to small proteins, restricting their usefulness. Specifically, the project seeks to streamline the design process by rapidly providing high-quality scaffolds to experimentalists while considerably reducing the time required for their creation. This rapidity is particularly important when swift actions are required for facing emerging and fast-evolving diseases. This will be achieved by employing recent advances in diffusion modelling, geometrical deep learning, and graph models and networks.