|
Abstract:
|
Single-cell RNA-seq data enable scientists to study cell developmental trajectories using various analysis pipelines, but these pipelines often lack a thorough statistical understanding. Here, we focus on studying the identifiability and convergence of embedding each cell into a lower dimensional space, a major component of most pipelines. Specifically, this embedding is performed with respect to a hierarchical model commonly used in single-cell analyses where the inner product between latent vectors is the natural parameter of an exponential family distributed random variable. Previous methods either use a less flexible statistical model or are too computationally intensive. We apply our method within a pipeline to analyze the oligodendrocytes in fetal mouse brains. Our results suggest that these cells mature into multiple different cell types, a result that coincides with other recent findings. This is in sharp contrast with more conventional analysis using less flexible models, which not only result in worse model diagnostics but also only suggest one mature cell type.
|
