Manifold Learning Uncovers Hidden Structure in Complex Cellular State Space

BIOENGINEERING SEMINAR
 
Abstract:
In the era of big biological data, there is a pressing need for methods that visualize, integrate and interpret high-throughput high-dimensional data to enable biological discovery. There are several major challenges in analyzing high-throughput biological data. These include the curse of (high) dimensionality, noise, sparsity, missing values, bias, and collection artifacts. In my work, I try to solve these problems using computational methods that are based on manifold learning. A manifold is a smoothly varying low-dimensional structure embedded within high-dimensional ambient measurement space. In my talk, I will present a number of recently completed and ongoing projects that utilize the manifold, implemented using graph signal processing and deep learning, to understand large biomedical datasets. First, I will present MAGIC, a data denoising and imputation method designed to ‘fix’ single-cell RNA-sequencing data. MAGIC uses data diffusion to learn the data manifold and at the same time fill in and smooth the data, thereby revealing the underlying structure of the data. I will show how MAGIC reveals a continuous phenotypic state-space in an epithelial-to-mesenchymal transition system. I will then talk about PHATE, a dimensionality reduction and visualization method specifically designed to reveal continuous progression structure. I will demonstrate how PHATE can give profound insight into a newly measured human embryonic stem cell system. Finally, I will talk about two deep learning methods that use specially designed constraints to allow for deep interpretable representations of heterogeneous systems such as gut microbiome data and single cell-data of tumor infiltrating lymphocytes.

Bio:
David van Dijk is an Associate Research Scientist in the lab of Smita Krishnaswamy at Yale University, departments of Genetics and Computer Science. His current research focuses on machine learning, and in specific manifold learning, applied to big biological data.
Previously, Dr. van Dijk worked with Eran Segal at the Weizmann Institute of Science, where he developed methods for predicting gene expression from DNA sequence.
He did his Ph.D. with Jaap Kaandorp in the Computational Science group at the University of Amsterdam, and with Eran Segal at the Weizmann Institute of Science in Rehovot, Israel.


Zoom link for attending remotely:  https://epfl.zoom.us/j/943730674