Division-resolved inference of flow and trajectories in proliferating cell populations

Abstract

High-throughput single-cell assays are widely used to quantify distributions of cell size, morphology, and molecular content across thousands of cells. However, such population distributions do not reveal how the measured cellular states change within individual cells over time. We introduce division-resolved inference of flow and trajectories (DRIFT), a computational framework that infers the dynamics of a measured cellular state from population distributions collected over time, without synchronizing or tracking individual cells. DRIFT solves a population-balance equation to separate state progression from the redistribution caused by cell division in proliferating populations. In simulations of growth and division perturbations, DRIFT recovered the ground-truth mean volume trajectories across simulated single-cell lineages. In live L1210 leukemia cells, DRIFT inferred perturbation-specific volume trajectories that were consistent with longitudinal single-cell measurements. Beyond cell volume, DRIFT also inferred DNA-content dynamics from fixed-cell flow cytometry in L1210 cells, consistent with independent DNA-synthesis assays. In live HeLa cells, DRIFT inferred cell area dynamics that were validated by continuous imaging. Overall, DRIFT converts endpoint measurements of cell populations into division-resolved cellular dynamics, providing a scalable strategy for high-throughput drug-response screening and mechanistic investigation.