Wenpin Hou, PhD
Postdoctoral Fellow,
Johns Hopkins University, Department of Biostatistics

From temporal gene expression to spatiotemporal gene regulation: statistical methods for analyzing single-cell genomic data

Pseudotime analysis with single-cell RNA-sequencing data has been widely used to study dynamic gene expression and regulatory programs in continuous biological processes.  While the number of studies with multiple single-cell samples is increasing dramatically, most pseudotime analysis methods ignore the sample-to-sample variability and cannot effectively handle the multi-sample data. We developed Lamian, a comprehensive and statistically rigorous computational framework to perform differential pseudotime analysis for multi-sample single-cell data. Lamian identifies temporal changes of gene expression and cell compositions associated with sample covariates, such as different biological conditions. Unlike existing methods that ignore sample-to-sample variability, Lamian draws statistical inference after accounting for cross-sample variability. Hence Lamian substantially reduces sample-specific false discoveries that are not generalizable to new samples and better characterizes the population patterns. We demonstrate the advantages of Lamian in decoding temporal gene expression programs through simulations and a real single-cell RNA-seq dataset studying COVID-19 immune response. We further developed a computational method to predict DNA methylation, a type of gene regulatory mechanism, using gene expression information in single cells and in a spatial context. We will extend Lamian to systematically identify temporal and spatial gene expression and regulatory programs in a mouse model of pancreatic disease.