Machine Learning for Single-Cell Sequencing Data Analysis

For the majority of complex diseases, specific subpopulations of cells underlying the disease are largely unknown. Identifying specific disease-relevant cells is crucial for understanding the biological mechanisms of diseases. Most current read-outs of sequence represent an average of all cells and do not resolve individual cells that are relevant to disease. Single-cell sequencing technology has advanced rapidly in the past few years, enabling us to study genomes, epigenomes, and transcriptomes of disease-relevant cell types at the single-cell resolution. This technology is revolutionizing our understanding of disease mechanisms. We develop machine learning methods for the analysis of single-cell sequencing data in order to identify disease-relevant cells and understand disease gene regulation at the single-cell level.

New advancements in biotechnology have enabled the generation of diverse types of omic data that can be used to answer many questions related to diseases at a scale not possible before. Due to the complexity of many disorders, including autoimmune disorders and different cancer types, a single type of data is not informative enough to capture all factors relevant to disease, and therefore, integrating multiple types of biological data is crucial. The main challenge in data integration is how to develop effective models that provide a comprehensive insight into the disease mechanisms. We develop computational, statistical and machine learning methods to integrate multiple types of large-scale biological data. A major aspect of our research focuses on autoimmune diseases, where we investigate mechanisms of gene regulation in autoimmune disease by integrating multi-omics data.