Mechanistic inference of stochastic gene expression from structured single-cell data

Single-cell gene expression measurements encode variability spanning molecular noise, cell-to-cell heterogeneity, and technical artifacts. Mechanistic stochastic models provide powerful approaches to disentangle these sources, yet inferring underlying dynamics from standard snapshot sequencing data faces fundamental identifiability limitations. This review focuses on how structured datasets with temporal, spatial, or multimodal features offer constraints to resolve these ambiguities, but demand more sophisticated models and inference strategies, including machine-learning techniques with inherent tradeoffs. We highlight recent progress in the judicious integration of structured single-cell data, stochastic model development, and innovative inference strategies to extract predictive, gene-level insights. These advances lay the foundation for scaling mechanistic inference upward to regulatory networks and multicellular tissues. • Mechanistic models reveal interpretable structure in expression variability. • Temporal, spatial, and multimodal data reshape what mechanisms can be inferred. • Increased data and model complexity challenge identifiability and reliability. • Simulation-based and ML methods expand inference, but require careful validation. • Principled single-gene approaches lay foundations for regulatory & multicellular systems.