Bayesian Models for Understanding Chromatin Structure in Single-cell Data
In recent years, single-cell technologies have revolutionized biology, providing exciting opportunities to map cellular populations through development. Among these technologies, single-cell ATAC-seq has become the leading assay for mapping the local structure of DNA, chromatin. How do we discover structure in this high-dimensional data and use it to understand cellular development? I have tackled this problem at the single-cell level, and aggregated populations. First, I will present Chroma, a Bayesian state-space model to characterize aggregated chromatin information by modeling the duration of functional and accessible chromatin regions. I will introduce hidden semi-Markov models as a biologically plausible assumption to distill regulatory regions from ATAC-seq data sets. Next, I will introduce MultiVI, a deep generative model for the joint analysis of chromatin and transcriptional information at the single-cell level. MultiVI offers a principled method to analyze both paired and unpaired samples jointly. Finally, I will show how these tools can be used to map the chromatin developmental landscape of cortical Interneurons.