Speaker
Description
The recent expansion of single-cell sequencing technologies has enabled simultaneous genome-wide measurements of multiple modalities in the same single cell. The potential to jointly profile such modalities as gene expression, chromatin accessibility, proteins, or multiple histone modifications at single-cell resolution represents a compelling opportunity to study biological processes at multiple layers of gene regulation. Analysis of single-cell multimodal datasets poses significant computational challenges because each single-cell data modality is high-dimensional. The number of measured features spans from hundreds in the case of protein epitopes to hundreds of thousands for chromatin-accessible sites. The multiple modalities profiled correspond to consecutive stages of gene expression, from its regulation by modifying chromatin architecture and engaging transcription-initiation proteins to the synthesis of mRNA and protein molecules. Thus, all modalities need to be modeled simultaneously to analyze and visualize multimodal data. I will present recently developed machine learning methods in our laboratory for (i) visualization and exploration of developmental processes with diffusion-based approaches, and (ii) inference of gene regulatory and expression programs leveraging topic modeling approaches.
