🔬 condensed matter

Single-cell-level distributions and relationships can differentiate cell-division and growth models

This paper demonstrates that analyzing single-cell-level probability distributions and statistical relationships of physiological quantities, such as birth and division sizes, enables the differentiation of both cell-division mechanisms (Timer, Sizer, Adder) and growth paradigms (linear, exponential), even in the presence of stochasticity, asymmetric partitioning, and lineage correlations, as validated by experimental data.

Original authors: Vikas, Rahul Marathe, Anjan Roy

Published 2026-07-02

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Vikas, Rahul Marathe, Anjan Roy

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a bustling city of tiny, single-celled organisms. Every time one of these cells splits in two, it's like a parent handing off a piece of their life to a child. But how do they decide when to split and how big to be? Do they split after a set amount of time, like a timer on a microwave? Do they wait until they reach a specific size, like a child waiting to reach a certain height before getting a haircut? Or do they simply grow a fixed amount of "stuff" (biomass) before splitting, regardless of how big they started?

This paper is like a detective story. The authors (Vikas, Marathe, and Roy) are trying to figure out which of these three "rules" (Timer, Sizer, or Adder) bacteria actually follow. They also want to know if the bacteria grow like a balloon inflating (exponentially) or like a car driving at a steady speed (linearly).

Here is the breakdown of their investigation using simple analogies:

The Three Rules of the Game

The Timer: Imagine a cell that has a strict alarm clock. No matter how big or small it is, it splits exactly 1 hour after it was born.
- The Problem: If a cell is born tiny, it will still be tiny when the alarm rings. If it's born huge, it stays huge. Over many generations, the family line would get either impossibly small or impossibly huge. The authors show that unless the timing is perfectly tuned (which is rare in nature), this rule leads to chaos. It's like a factory that produces cars of random sizes because the assembly line runs for a fixed time regardless of the parts available.
The Sizer: Imagine a cell that has a "size limit" sign. It keeps growing until it hits a specific weight (say, 2 grams), and then it splits.
- The Result: If a baby cell is born small, it has to grow a lot to hit the sign. If it's born big, it doesn't have to grow much. This naturally keeps the family size stable. It's like a school that only lets students graduate when they reach a specific height; short kids grow more, tall kids grow less, and everyone ends up the same height eventually.
The Adder: Imagine a cell that says, "I don't care how big I am when I was born. I just need to add exactly 1 gram of new stuff to my body before I split."
- The Result: This also keeps the family size stable. If you start small, you add 1 gram and split. If you start big, you add 1 gram and split. It's like a savings account where you only withdraw money once you've deposited a fixed amount, regardless of your current balance.

The Detective Work: How to Tell Them Apart

The authors realized that just looking at the average size or time isn't enough to tell these rules apart. It's like trying to guess a person's personality just by knowing their average height; you need more detail.

Instead, they looked at the entire distribution (the full range of sizes and times) and the relationships between different measurements. Think of it like this:

The "Fingerprint" Approach: They calculated the mathematical "shape" of the data for each model. If you have a jar of marbles (cells) and you measure their sizes, the pattern of sizes for a "Sizer" population looks different from an "Adder" population, even if their averages are the same.
The "Correlation" Approach: They looked at how different traits talk to each other. For example, does a cell that is born small tend to grow for a longer time?
- In the Sizer model, a small birth size means a long growth time (to reach the target size).
- In the Adder model, a small birth size doesn't necessarily mean a longer growth time in the same way, because the goal is just to add a fixed amount.

By comparing these mathematical "fingerprints" and "conversations" between traits against real experimental data (from bacteria like E. coli and M. tuberculosis), they could see which rule the bacteria were actually following.

The Big Findings

Bacteria are "Adders": When they looked at real data from E. coli and M. tuberculosis, the patterns matched the Adder model best. The bacteria aren't waiting for a timer or a specific size; they are simply growing a fixed amount of "new stuff" before dividing.
Exponential Growth: They also checked if the bacteria were growing like a balloon (exponentially) or a car (linearly). The data suggested that E. coli grows exponentially (getting bigger faster as it gets bigger), while for M. tuberculosis, it was harder to tell the difference between linear and exponential growth, but the "Adder" rule still held true.
Robustness: Even if the growth rate of a parent cell is similar to its grandparents (a "family trait"), the math still works. The rules they derived remain valid even with these family connections.

Why This Matters (According to the Paper)

The paper argues that by understanding these simple statistical patterns, scientists can figure out the complex molecular "machinery" inside the cell without needing to see every single molecule. It's like figuring out how a car engine works just by listening to the sound of the exhaust and measuring the speed, rather than taking the engine apart.

If we know a bacterium follows the "Adder" rule, we know we should look for molecular mechanisms that count "added biomass" rather than mechanisms that measure "total size" or "time." This helps us understand the evolutionary strategies of different microbes and how they keep their sizes stable to survive.

In short: The authors built a mathematical toolkit to listen to the "conversations" between cell sizes and growth times. By doing so, they proved that many bacteria follow a "grow-a-fixed-amount" strategy (the Adder) and grow exponentially, providing a clear way to identify these biological rules in the wild.

Technical Summary: Single-cell-level distributions and relationships can differentiate cell-division and growth models

Problem Statement
Understanding the regulatory mechanisms governing bacterial cell growth and division is fundamental to microbiology and biotechnology. While it is established that cell division follows specific strategies—historically categorized as Timer (division by age), Sizer (division by critical size), or Adder (division after adding constant biomass)—distinguishing between these models and identifying the underlying single-cell growth paradigm (linear vs. exponential) remains challenging. Existing methods primarily rely on correlations between physiological quantities (birth-size, division-size, division-time, and division-added-size). However, these correlations are often derived under idealized assumptions (e.g., symmetric, non-stochastic partitioning and deterministic growth rates). Furthermore, experimental data often suffer from insufficient precision to directly determine growth types, necessitating indirect differentiation methods. The authors address the need for a comprehensive analytical framework that accounts for stochastic growth rates, asymmetric partitioning, and lineage autocorrelations to robustly differentiate between division and growth models.

Methodology
The study employs a coarse-grained analytical approach to derive probability distributions and statistical relationships for single-cell quantities. The model is built upon three free-parameter distributions:

Principal Distribution: Determines the division rule (division-time $\tau_d$ for Timer, division-size $s_d$ for Sizer, or division-added-size $\Delta_d$ for Adder).
Growth-rate Distribution ( $\Lambda(\alpha)$ ): Accounts for stochasticity in growth rates across a lineage.
Partitioning-fraction Distribution ( $\kappa(\beta)$ ): Accounts for stochasticity and asymmetry in the inheritance of size by daughter cells.

Using probability transformations and statistical arguments (detailed in Appendices A and B), the authors analytically derive the distributions for all other single-cell quantities ( $s_b, s_d, \tau_d, \Delta_d$ ) for both Sizer and Adder models under linear and exponential growth regimes. The Timer model is excluded from the primary analysis of homeostasis under exponential growth, as it fails to maintain size stability without fine-tuned parameters.

The analytical results are validated against published experimental data from microfluidic studies (Mother Machine setups) involving Escherichia coli and Mycobacterium tuberculosis. The authors fit Johnson family distributions to the experimental data for the free parameters and compare the resulting analytical predictions for single-cell distributions and statistical relationships against the experimental observations.

Key Contributions

Extension to Stochasticity and Asymmetry: The study extends existing correlation-based differentiation methods to include stochastic growth rates and asymmetric, stochastic partitioning, providing a more biologically realistic framework.
Analytical Distributions: The authors provide closed-form analytical expressions for the probability distributions of birth-size, division-size, division-time, and division-added-size for Sizer and Adder models under both linear and exponential growth.
Statistical Relationships: The paper derives specific relationships between the means, standard deviations, and coefficients of variation of these quantities. Crucially, it identifies that while mean-based relationships are often universal across models, relationships involving standard deviations and correlations can uniquely distinguish between division models (Sizer vs. Adder) and growth types (linear vs. exponential).
Robustness to Autocorrelations: The analysis demonstrates that the derived results remain valid even when the growth rate of a cell is correlated with the growth rates of its ancestors (lineage autocorrelation), a phenomenon observed in recent experimental data.

Results

Differentiation via Distributions: By comparing analytical predictions with experimental data for E. coli (at 25°C, 27°C, and 37°C) and M. tuberculosis, the authors found that the exponential Adder model provided the closest agreement with the observed distributions of single-cell quantities. The Sizer model and linear growth models showed poorer fits.
Differentiation via Standard Deviations: The relationship between the standard deviation of division-added-size ( $\sigma_{\Delta_d}$ ) and division-size ( $\sigma_{s_d}$ ) successfully distinguished between Sizer and Adder models. Experimental data points clustered closer to the Adder prediction.
Differentiation via Correlations: The correlation between division-time ( $\tau_d$ ) and birth-size ( $s_b$ ) was shown to depend on both the division model and the growth type. Experimental correlations for E. coli aligned best with the exponential Adder model.
Growth Type Ambiguity: For M. tuberculosis, the data suggested an Adder division strategy, but the distinction between linear and exponential growth was less clear, with both models fitting the data similarly well.
Homeostasis: The study confirms that Sizer and Adder models naturally maintain cell-size homeostasis, whereas the Timer model generally does not under exponential growth unless parameters are perfectly tuned, leading to divergence in birth-size over generations.

Significance
The paper asserts that single-cell-level distributions and the statistical relationships between them serve as powerful tools to differentiate not only between cell-division strategies (Sizer vs. Adder) but also between single-cell growth paradigms (linear vs. exponential). This capability allows researchers to pinpoint the underlying regulatory mechanisms controlling cell size and division in various microbial species without needing to resolve the specific molecular networks immediately. The findings aid in identifying evolutionary strategies or constraints in different species and provide a robust analytical basis for interpreting single-cell experimental data, even in the presence of stochasticity and lineage correlations. The authors note that while their coarse-grained approach is robust to molecular details, future work could extend these analyses to mixed or more complex growth models (e.g., bilinear or super-exponential) to further improve agreement with experimental data.

The Three Rules of the Game

The Detective Work: How to Tell Them Apart

The Big Findings

Why This Matters (According to the Paper)

More like this