Reconstructing developmental trajectories using latent dynamical systems and time-resolved transcriptomics

• DOI: 10.1016/j.cels.2024.04.004
• Publication Date: 2024-05-15T14:41:23Z
• Authors: Rory J. Maizels; Daniel M. Snell; James Briscoe
• Keywords: Snapshot (computer storage); Autoencoder; Cell fate determination

The snapshot nature of single-cell transcriptomics presents a challenge for studying the dynamics cell fate decisions. Metabolic labeling and splicing can provide temporal information at level, but current methods have limitations. Here, we present framework that overcomes these limitations: experimentally, developed sci-FATE2, an optimized method metabolic with increased data quality, which used to profile 45,000 embryonic stem (ES) cells differentiating into neural tube identities. Computationally, two-stage dynamical modeling: VelvetVAE, variational autoencoder (VAE) velocity inference outperforms all other tools tested, VelvetSDE, stochastic differential equation (nSDE) simulating trajectory distributions. These recapitulate underlying dataset distributions capture features such as decision boundaries between alternative fates fate-specific gene expression. recast analyses from descriptions observed models generated them, providing investigating developmental