Graph Representation Learning in Biology

Catalog description

Biological and cellular systems are often modeled as graphs (networks) of interacting elements. This approach has been highly successful owing to the theory, methodology and algorithms that support analysis and learning on graphs. However, recent advances in deep learning techniques have led to a surge in research on graph representation learning. In particular, these advances have led to new state-of-the-art results in biomedicine and healthcare. This course will provide a synthesis and overview of graph representation learning within systems biology and medicine. We will begin with a discussion of the goals of graph representation learning, as well as key methodological foundations in graph theory and network analysis. We next review traditional methods such as topological descriptors, graph kernels, and spectral graph theory. We then introduce techniques for learning node embeddings, including random-walk based methods and applications to knowledge graphs. We finally provide a technical synthesis and introduction to the highly successful graph neural network formalism as well as recent advancements in deep generative models for graphs.