Dynamic distribution decomposition for single-cell snapshot time series identifies subpopulations and trajectories during iPSC reprogramming.

摘要: Recent high-dimensional single-cell technologies such as mass cytometry are enabling time series experiments to monitor the temporal evolution of cell state distributions and identify dynamically important states, fate decision states in differentiation. However, these destructive, require analysis approaches that temporally map between across points. Current approximate a dynamical system suffer from too restrictive assumptions about type kinetics, or link together pairs sequential measurements discontinuous fashion. We propose Dynamic Distribution Decomposition (DDD), an operator approximation approach infer continuous distribution On basis snapshot data, DDD approximates Perron-Frobenius by means finite set functions. This procedure can be interpreted Markov chain over continuum states. By only assuming memoryless (autonomous) process, types dynamics represented more general than those other common models, e.g., chemical reaction networks, stochastic differential equations. Furthermore, we posteriori check whether autonomy valid calculation prediction error—which show gives measure within studied system. The continuity ensure same maps all points, not arbitrarily changing at each point. demonstrate ability reconstruct their transitions both on synthetic well iPSC reprogramming fibroblast use find previously identified subpopulations cells visualise differentiation trajectories. allows interpretation data low-dimensional thereby interpretable for variety biological processes identifying