Pooling single-cell recordings: Scalable inference through heterogeneous kinetics

TL;DR

This paper presents a scalable Bayesian inference framework for analyzing pooled single-cell data, effectively accounting for cell-to-cell variability to reconstruct intracellular processes and infer kinetic parameters.

Contribution

The authors introduce an exact Bayesian method that models heterogeneous kinetics in pooled single-cell measurements, improving process reconstruction and parameter inference.

Findings

Successfully reconstructed gene expression dynamics in yeast.

Found no evidence of a refractory period in GAL1 promoter.

Method effectively separates intrinsic, extrinsic, and technical variability.

Abstract

Mathematical methods together with measurements of single-cell dynamics provide unprecedented means to reconstruct intracellular processes that are only partly or indirectly accessible experimentally. To obtain reliable reconstructions the pooling of measurements from several cells of a clonal population is mandatory. The population's considerable cell-to-cell variability originating from diverse sources poses novel computational challenges for process reconstruction. We introduce an exact Bayesian inference framework that properly accounts for the population heterogeneity but also retains scalability with respect to the number of pooled cells. The key ingredient is a stochastic process that captures the heterogeneous kinetics of a population. The method allows to infer inaccessible molecular states, kinetic parameters, compute Bayes factors and to dissect intrinsic, extrinsic and technical contributions to the variability in the data. We also show how additional single-cell readouts such as morphological features can be included into the analysis. We then reconstruct the expression dynamics of a gene under an inducible GAL1 promoter in yeast from time-lapse microscopy data. Based on Bayesian model selection the data yields no evidence of a refractory period for this promoter.