I'm Duncan, a Postdoc in the Department of Molecular Genetics at The University of Toronto.

I work on integrating the plethora of diverse biological datasets to develop a unified set of gene/protein function annotations and accurately describe cellular functional organization at scale. Good quality gene function descriptions are crucial for understanding how the cell works. Single datasets on their own are often limited in coverage of the genome, have unavoidable biases due to the underlying experimental method, and contain noise. By integrating datasets, we can increase coverage, control bias, and reduce noise.

I've developed a deep learning network integration algorithm, BIONIC, which uses graph convolutional networks (GCNs) to learn topological features for genes and proteins across different interaction networks (protein-protein interaction, coexpression, genetic interaction, etc.). These network-specific features are combined to generate integrated gene and protein features, which encode functional information from the input biological networks. These integrated features can then be clustered to identify groups of functionally related genes/proteins (i.e. functional modules), potentially identifying new subsystems in the cell.

If you'd like to get in touch, you can email me at duncan.forster@mail.utoronto.ca.