Phenotypic selection occurs when genetically identical cells are subject to different reproductive abilities due to cellular noise. Such noise arises from fluctuations in reactions synthesizing proteins and plays a crucial role in how cells make decisions and respond to stress or drugs. We propose a general stochastic agent-based model for growing populations capturing the feedback between gene expression and cell division dynamics. We devise a finite state projection approach to analyze gene expression and division distributions and infer selection from single-cell data in mother machines and lineage trees. We use the theory to quantify selection in multi-stable gene expression networks and elucidate that the trade-off between phenotypic switching and selection enables robust decision-making essential for synthetic circuits and developmental lineage decisions. Using live-cell data, we demonstrate that combining theory and inference provides quantitative insights into bet-hedging–like response to DNA damage and adaptation during antibiotic exposure in Escherichia coli.

Stochastic models of gene expression are routinely used to explain large variability in measured mRNA levels between cells. These models typically predict the distribution of the total mRNA level per cell but ignore compartment-specific measurements which are becoming increasingly common. Here we construct a two-compartment model that describes promoter switching between active and inactive states, transcription of nuclear mRNA and its export to the cytoplasm where it decays. We obtain an analytical solution for the joint distribution of nuclear and cytoplasmic mRNA levels in steady-state conditions. Based on this solution, we build an efficient and accurate parameter inference method which is orders of magnitude faster than conventional methods. If you want to participate in the seminar, you need to enter IBS builiding (https://www.ibs.re.kr/bimag/visiting/). Please contact if you first come IBS to get permission to enter IBS building.