Seminar: Decoding Tissue and Cellular Complexity using Network Models and Deep Learning

Abstract

Biological systems are sustained by complex communication and coordination among various types of molecular signals and pathways. Each type of molecule (e.g., DNA, proteins and chemicals) is profiled separately and falls into different data modalities that capture unique snapshots of cell state and function. Although improvements in sequencing technologies now cover more depth and breadth of this space, individual modalities only capture partial, sparse, and static snapshots of dynamic molecular processes. In addition, technical and ethical issues restrict the types of tissues and cells that can be isolated or studied directly in humans. Because of this, our knowledge of the complex biological circuitry within tissues and cells remains largely obscure.

In this talk, I will present several efforts to fill the knowledge gap using network modeling and deep learning-based frameworks. First, I will briefly introduce my work using tissue/cell type-specific networks to predict genes underlying complex human diseases. Then, I will present novel semi-supervised deep learning frameworks for obtaining multimodal and dynamic views of cells, and transferring single-cell knowledge across species. The models I developed have been used to identify novel autism genes, reveal sex differences during development, characterize gene evolution, and predict human Alzheimer’s disease related genes based on mouse models. Finally, I will discuss several exciting opportunities I envision for using machine learning to model how cells change throughout development, with disease, and in reaction to various stimuli such as viruses. In doing so, I aim to provide generalizable tools for the community to characterize multimodal pathways altered in diseases that can ultimately guide disease diagnosis and intervention.

Biography

Dr. Ran Zhang is currently a Data Science Postdoctoral Fellow at the University of Washington, working with Dr. William Noble building integrative models to study single cells. Before that, she received a PhD from Princeton University, working with Dr. Olga Troyanskaya building functional networks to study developmental and neurodegenerative diseases. Her research lies at the intersection of computer science and biology, with a focus on leveraging public data with machine learning to generate data-driven hypotheses when there are only partial observations.