From pseudotime to true dynamics: reconstructing a real-time axis for T cells differentiation

Abstract

Numerous methods have recently emerged for ordering single cells along developmental trajectories. However, accurate depiction of developmental dynamics can only be achieved after rescaling the trajectory according to the relative time spent at each developmental point. We formulate a model which estimates local cell densities and fluxes, and incorporates cell division and apoptosis rates, to infer the real time dimension of the developmental trajectory. We validate the model using mathematical simulations, and apply it on experimental high dimensional cytometry data obtained from the mouse thymus to construct the true time-profile of the thymocyte developmental process. Our method can easily be implemented in any of the existing tools for trajectory inference.