arxiv: 2604.26082 · v1 · submitted 2026-04-28 · 🧬 q-bio.PE

Recognition: unknown

How is gene-regulatory evolution affected by cell-to-cell variability?

James Holehouse, Leonardo Ivan Estrella Dzib

Authors on Pith no claims yet

Pith reviewed 2026-05-07 13:41 UTC · model grok-4.3

classification 🧬 q-bio.PE

keywords gene-regulatory networkscell-to-cell variabilitygene expression noiseevolutionary dynamicsnetwork alignmentfeedforward loopsnetwork robustness

0 comments

The pith

Increased cell-to-cell variability selects for aligned and robust gene-regulatory networks enriched in specific motifs.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper extends a classical model of gene-regulatory network evolution to incorporate noisy developmental dynamics, allowing one network to produce a distribution of phenotypes. It defines an alignment score, drawing from Hopfield network ideas, that measures how signed gene interactions reinforce a target stable state. Simulations show that higher noise levels drive populations toward networks with greater alignment, shorter developmental paths, and optimized fitness. These aligned networks are enriched for coherent feedforward and positive feedback loops and resist mutational changes better than noiseless or non-evolved controls. Alignment works by creating redundancy in reinforcing interactions, turning variability into a selective pressure that favors robustness.

Core claim

When developmental dynamics include noise, selection favors networks in which a larger fraction of gene-gene interactions carry the correct sign to support the target phenotype. This alignment score rises with noise, producing architectures that contain more coherent feedforward loops and positive feedback loops, maintain fitness while shortening paths to stability, and lose less performance when mutated.

What carries the argument

The alignment score that counts the net reinforcing gene-gene interactions supporting a stable target phenotype.

If this is right

Evolved networks reduce the number of steps needed to reach a stable phenotype.
Higher noise increases the fraction of appropriately signed interactions and enriches coherent feedforward and positive feedback motifs.
Aligned networks maintain fitness after random mutations better than non-aligned controls.
Populations under noise reach higher fitness values than equivalent noiseless populations.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Noise may act as a constructive evolutionary force that favors redundancy over precision in regulatory wiring.
Empirical gene-regulatory networks from cells with high expression variability could be scored for alignment to test whether natural selection produces similar patterns.
The model suggests that removing noise experimentally might destabilize certain regulatory motifs that evolved under variable conditions.

Load-bearing premise

The alignment score meaningfully measures fitness under noise and the extended model captures the selective pressures that act on real gene-regulatory networks.

What would settle it

Running the same evolutionary simulations with noise but without rewarding alignment, then checking whether motif enrichment and mutational buffering still appear at the same rate.

Figures

Figures reproduced from arXiv: 2604.26082 by James Holehouse, Leonardo Ivan Estrella Dzib.

**Figure 1.** Figure 1: Summary of the findings of our study. (A) Stochasticity in the regulatory connections comprising the genotype can lead to a diverse set of phenotypes. (B) GRNs evolved in stochastic environments develop mechanisms to express optimal phenotypes even in highly variable environments. (C) Successful GRNs become highly aligned with the optimal phenotype, meaning that very few regulatory connections lead to devi… view at source ↗

**Figure 2.** Figure 2: Overview of the GRN evolution model used in this study. (c) The model studied in the main text builds on refs. [32, 33] in which the iterative equation shown constitutes the developmental dynamics that determine the eventual phenotype expressed by a given genotype. Between evolutionary time steps, elements of the genotype may mutate. (b)–(c) In our adaptation of the model from refs. [32, 33], we assume tha… view at source ↗

**Figure 3.** Figure 3: Examples of motifs in some aligned and non-aligned network motifs. The blue circles represent a positive and the red squares a negative entry in the target phenotype Popt. Positive edges (promoting interactions) are represented with a blue arrow, and negative ones (inhibitory interactions) with a squared red tip. By changing the direction of one edge (marked with a dashed line and an exclamation mark), we … view at source ↗

**Figure 4.** Figure 4: (A) Evolution of the average population fitness and (B) average mean path length per generation, across 30 independent replicates at different noise variances θ. Shaded regions represent 95% confidence intervals around the mean in the zoomed-out panels and 68% in the zoomed-in panels. Lighter colors correspond to smaller noise variances. The noiseless case (θ = 0) is shown with a dotted line. In panel (B),… view at source ↗

**Figure 5.** Figure 5: Evolution of the average alignment scores. (A) Average alignment score across 30 populations over generations. Each line represents the average alignment score for a population, and the color fillings represent 95% confidence intervals. Lighter colors represent smaller noise variances. The noiseless scenario is plotted with a dotted line. (B) Average alignment scores after evolution from 30 evolved populat… view at source ↗

**Figure 6.** Figure 6: Enrichment of network motifs as a result of noiseless evolution. Comparison of 30 populations evolved without noise (darker hues) against non-evolved populations of stable matrices (lighter hues). (A) Average concentration per type of FFL. On the x-axis, we place the type of loop with a diagram. The error bars represent 95% confidence intervals around the averages. (B) Average concentration of positive (gr… view at source ↗

**Figure 7.** Figure 7: Enrichment of network motifs as a result of noisy evolution, compared to noiseless evolution. Comparison of 30 populations evolved with high noise variance (θ = 8; darker hues) and no noise (lighter hues). (A) Average concentration per type of FFL. On the x-axis, we place the type of loop with a diagram. The error bars represent 95% confidence intervals around the averages. (B) Average concentration of pos… view at source ↗

**Figure 8.** Figure 8: Change in mutational robustness across evolved populations before and after evolution. (A) Change in average stable expression shift, (B) average stable expression variance, (C) average instability difference, and (D) instability variance after the evolutionary processes for 30 populations across various noise levels (plotted as standard deviations in the x-axis). We fix the noise variance to θfixed = 1 to… view at source ↗

read the original abstract

The evolutionary origins of structural features in reconstructed gene-regulatory networks (GRNs) remain poorly understood, especially given the random aspects of gene expression. Here, we extend a classical model of GRN evolution to allow a single network to express a distribution of phenotypes through noisy developmental dynamics. Inspired by Hopfield networks, we introduce an alignment score that quantifies the cohesion of gene-gene interactions in the network to support a target stable phenotype. Overall, evolved populations optimized their fitness and reduced the length of their developmental paths. Increased noise levels promoted alignment, enriched coherent feedforward and positive feedback loops relative to non-evolved and noiseless controls, and buffered against mutational perturbations. Alignment provides intuitive interpretations because an increased number of appropriately signed gene-gene interactions is more redundant and thus more robust against developmental noise and mutations. Together, these results demonstrate that cell-to-cell variability exerts strong selective pressure, driving the evolution of aligned, robust, and motif-enriched GRN architectures.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Noise selects for aligned, motif-enriched, and mutation-robust GRNs in these evolutionary simulations.

read the letter

The key point is that cell-to-cell noise in this model selects for GRNs that are better aligned to their target phenotype, enriched for certain motifs, and more robust to mutations. The authors extend an existing GRN evolution framework by incorporating noisy developmental dynamics, allowing one network to express a distribution of phenotypes. They introduce an alignment score inspired by Hopfield fixed-point stability to quantify how well the interaction signs support a stable state. Fitness is tied to phenotype reachability under noise. Their simulations show that higher noise promotes higher alignment scores, increases the frequency of coherent feedforward loops and positive feedback loops compared to controls, and improves resistance to mutational changes. The intuition they offer is that aligned networks have redundant signed interactions that stabilize the phenotype against perturbations. This is a straightforward and useful addition to the literature on GRN evolution. The results are consistent with the idea that variability acts as a selective force favoring robustness, and the motif findings build on known work without overclaiming. The paper does a good job of linking the alignment metric to both noise buffering and mutational robustness in an intuitive way. The main limitations are that everything is simulation-based with no direct comparison to real GRN data, and the alignment score is constructed relative to the target, which could make some interpretations somewhat dependent on the fitness definition. Without seeing the full methods, it is difficult to assess how thoroughly they ruled out simulation-specific artifacts or tested parameter sensitivity. These are typical for this type of computational study rather than fatal issues. This work is for researchers in evolutionary systems biology and synthetic biology who think about noise in gene regulation. A reader interested in how stochasticity influences network architecture would get value from the simulations and the proposed mechanism. It is solid enough to warrant peer review, as the central claims can be evaluated on the simulation outputs and the model is an incremental but clear extension.

Referee Report

3 major / 2 minor

Summary. The paper extends a classical model of gene-regulatory network (GRN) evolution to incorporate cell-to-cell variability through noisy developmental dynamics. It introduces an alignment score inspired by Hopfield networks to quantify the cohesion of gene-gene interactions supporting a target stable phenotype. Evolutionary simulations are used to show that higher noise levels increase alignment, enrich coherent feedforward loops and positive feedback loops relative to non-evolved and noiseless controls, reduce developmental path lengths, and buffer against mutational perturbations. The authors conclude that noise exerts selective pressure favoring aligned, robust, and motif-enriched GRN architectures.

Significance. If the simulation results prove robust upon full methodological disclosure, this would be a significant contribution to evolutionary systems biology by providing a mechanistic link between gene expression noise and the evolution of GRN motifs and robustness. The use of explicit controls (non-evolved and noiseless) and the intuitive redundancy interpretation of alignment are strengths that allow isolation of noise effects. The work addresses a gap in understanding how random aspects of gene expression shape network evolution.

major comments (3)

Methods section on the extended model: The implementation of noisy developmental dynamics lacks sufficient detail on the noise model (additive vs. multiplicative), specific variance parameters, sampling of phenotype distributions, and controls for simulation artifacts. This is load-bearing for the central claim, as the reported promotion of alignment and motif enrichment under increased noise cannot be evaluated without these to confirm robustness against artifacts.
Section introducing the alignment score and fitness definition: The alignment score is defined relative to the target phenotype (inspired by Hopfield fixed-point stability), and fitness depends on reaching that phenotype under noisy dynamics. An explicit mathematical statement is needed showing that the evolutionary process does not favor alignment by construction, to address the partial dependence noted in the alignment-fitness relationship.
Results on motif enrichment and mutational buffering: The claims of enrichment for coherent feedforward and positive feedback loops, plus buffering against perturbations, are presented without statistical tests (e.g., p-values, confidence intervals, or details on the null model for motif counting). This undermines assessment of whether the effects are significant or merely descriptive.

minor comments (2)

Ensure all simulation parameters (population size, mutation rates, noise levels, generations) are listed explicitly, ideally in a table, for reproducibility.
Clarify the specific classical GRN evolution model being extended, with a direct citation to the original reference in the introduction.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive and detailed comments, which have identified important areas for improving the clarity, rigor, and reproducibility of our work. We have revised the manuscript accordingly and address each major comment below.

read point-by-point responses

Referee: Methods section on the extended model: The implementation of noisy developmental dynamics lacks sufficient detail on the noise model (additive vs. multiplicative), specific variance parameters, sampling of phenotype distributions, and controls for simulation artifacts. This is load-bearing for the central claim, as the reported promotion of alignment and motif enrichment under increased noise cannot be evaluated without these to confirm robustness against artifacts.

Authors: We agree that the Methods section requires expanded detail for full reproducibility and evaluation. In the revised manuscript, we now explicitly state that the noise is implemented as additive Gaussian perturbations to the gene expression update rule (with variance parameter σ² varied from 0.01 to 0.1 across conditions), that phenotype distributions are obtained by averaging 100 independent stochastic trajectories per network (each integrated to steady state), and that we include controls such as fixed random seeds across conditions, verification against multiplicative noise alternatives, and checks that results are insensitive to numerical integration tolerances. These additions directly address potential artifacts and support the robustness of the reported noise effects. revision: yes
Referee: Section introducing the alignment score and fitness definition: The alignment score is defined relative to the target phenotype (inspired by Hopfield fixed-point stability), and fitness depends on reaching that phenotype under noisy dynamics. An explicit mathematical statement is needed showing that the evolutionary process does not favor alignment by construction, to address the partial dependence noted in the alignment-fitness relationship.

Authors: We acknowledge the value of an explicit demonstration here. The revised manuscript now includes a dedicated subsection with the mathematical relationship: alignment A is the normalized sum of signed interactions consistent with the target phenotype vector (A = (1/|E|) Σ_{ij} w_{ij} s_i s_j), while fitness F is the empirical probability of reaching the target under noisy dynamics (F = fraction of trajectories with ||x_final - s|| < ε). We derive that ∂F/∂A is not identity and provide a counterexample of networks with high F but low A (sparse strong edges). We also add a control analysis of non-evolved networks showing that the observed increase in A under evolution exceeds the correlation expected from F alone, confirming alignment is not favored purely by construction. revision: yes
Referee: Results on motif enrichment and mutational buffering: The claims of enrichment for coherent feedforward and positive feedback loops, plus buffering against perturbations, are presented without statistical tests (e.g., p-values, confidence intervals, or details on the null model for motif counting). This undermines assessment of whether the effects are significant or merely descriptive.

Authors: We agree that formal statistical support strengthens these claims. The revised Results section now reports two-sided Wilcoxon rank-sum tests (with exact p-values) comparing motif frequencies in noisy-evolved networks against both noiseless-evolved and non-evolved controls, using 1000 degree-preserving randomizations as the null model for motif detection. For mutational buffering, we provide 95% bootstrap confidence intervals over 50 independent evolutionary replicates and paired statistical tests (p < 0.01 for all key enrichments and robustness differences). These additions confirm the effects are statistically significant rather than descriptive. revision: yes

Circularity Check

0 steps flagged

No significant circularity; results derive from independent simulation comparisons

full rationale

The paper's derivation consists of evolutionary simulations in which fitness is defined directly by reachability of a target phenotype under noisy developmental dynamics. The alignment score is introduced separately as an interpretive post-hoc metric (inspired by Hopfield fixed-point stability) that quantifies signed interaction cohesion for the same target; it is not inserted into the fitness function, selection rule, or any optimization step. Enrichment of motifs and mutational buffering are reported as direct outputs of comparing noisy versus noiseless evolved populations against non-evolved controls. No equation reduces to its own input by construction, no parameter is fitted and then relabeled as a prediction, and no load-bearing premise rests on a self-citation chain. The framework is therefore self-contained against external benchmarks of simulation reproducibility.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

Central claim depends on the validity of the alignment score as a fitness proxy and the assumption that the base GRN evolution model plus added noise captures biologically relevant dynamics; no free parameters or invented entities beyond the alignment score are described.

axioms (1)

domain assumption Classical models of GRN evolution remain valid when extended with noisy developmental dynamics.
Paper builds directly on prior models without re-deriving their foundations.

invented entities (1)

Alignment score no independent evidence
purpose: Quantifies cohesion of gene-gene interactions to support a target stable phenotype.
Newly defined metric inspired by Hopfield networks; no independent biological validation mentioned.

pith-pipeline@v0.9.0 · 5460 in / 1234 out tokens · 65297 ms · 2026-05-07T13:41:21.556156+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

62 extracted references · 41 canonical work pages

[1]

Modeling transcriptional regulatory networks

Hamid Bolouri and Eric H Davidson. “Modeling transcriptional regulatory networks”. In:BioEssays24.12 (2002), pp. 1118–1129

2002
[2]

The evolution of hierarchical gene regulatory networks

Douglas H Erwin and Eric H Davidson. “The evolution of hierarchical gene regulatory networks”. In:Nature Reviews Genetics10.2 (2009), pp. 141–148

2009
[3]

Emerging properties of animal gene regulatory networks

Eric H Davidson. “Emerging properties of animal gene regulatory networks”. In:Nature468.7326 (2010), pp. 911–920

2010
[4]

Employing core regulatory circuits to define cell identity

Nathalia Almeida et al. “Employing core regulatory circuits to define cell identity”. EN. In:The EMBO Journal (May 2021).DOI: 10.15252/embj.2020106785 .URL: https://www.embopress.org/doi/10.15252/ embj.2020106785

work page doi:10.15252/embj.2020106785 2021
[5]

Gene regulatory network reconstruction: harnessing the power of single-cell multi-omic data

Daniel Kim et al. “Gene regulatory network reconstruction: harnessing the power of single-cell multi-omic data”. en. In:npj Systems Biology and Applications9.1 (Oct. 2023). Publisher: Nature Publishing Group, p. 51.ISSN: 2056-7189.DOI: 10.1038/s41540-023-00312-6 .URL: https://www.nature.com/articles/s41540- 023-00312-6. 14

work page doi:10.1038/s41540-023-00312-6 2023
[6]

Gene regulatory network inference: evaluation and application to ovarian cancer allows the prioritization of drug targets

Piyush B. Madhamshettiwar et al. “Gene regulatory network inference: evaluation and application to ovarian cancer allows the prioritization of drug targets”. In:Genome Medicine4.5 (May 2012), p. 41.ISSN: 1756-994X. DOI:10.1186/gm340.URL:https://doi.org/10.1186/gm340

work page doi:10.1186/gm340.url:https://doi.org/10.1186/gm340 2012
[7]

A gene regulatory network controlling the embryonic specification of endoderm

Isabelle S Peter and Eric H Davidson. “A gene regulatory network controlling the embryonic specification of endoderm”. In:Nature474.7353 (2011), pp. 635–639

2011
[8]

Improving gene regulatory network inference and assessment: The importance of using network structure

Juan M. Escorcia-Rodríguez et al. “Improving gene regulatory network inference and assessment: The importance of using network structure”. en. In:Front. Genet.14 (Feb. 2023), p. 1143382.ISSN: 1664-8021.DOI: 10.3389/ fgene . 2023 . 1143382.URL: https : / / www . frontiersin . org / articles / 10 . 3389 / fgene . 2023 . 1143382/full

2023
[9]

System Principles Governing the Organization, Architecture, Dynamics, and Evolution of Gene Regulatory Networks

Julio A. Freyre-González et al. “System Principles Governing the Organization, Architecture, Dynamics, and Evolution of Gene Regulatory Networks”. English. In:Front. Bioeng. Biotechnol.10 (May 2022). Publisher: Frontiers.ISSN: 2296-4185.DOI: 10.3389/fbioe.2022.888732.URL: https://www.frontiersin.org/ journals/bioengineering-and-biotechnology/articles/10.3...

work page doi:10.3389/fbioe.2022.888732.url: 2022
[10]

A comprehensive overview and critical evaluation of gene regulatory network inference technologies

Mengyuan Zhao et al. “A comprehensive overview and critical evaluation of gene regulatory network inference technologies”. In:Briefings in Bioinformatics22.5 (Sept. 2021), bbab009.ISSN: 1477-4054.DOI: 10.1093/ bib/bbab009.URL:https://doi.org/10.1093/bib/bbab009

work page doi:10.1093/bib/bbab009 2021
[11]

Advantages and limitations of current network inference methods

Riet De Smet and Kathleen Marchal. “Advantages and limitations of current network inference methods”. en. In: Nat Rev Microbiol8.10 (Oct. 2010), pp. 717–729.ISSN: 1740-1526, 1740-1534.DOI: 10.1038/nrmicro2419. URL:https://www.nature.com/articles/nrmicro2419

work page doi:10.1038/nrmicro2419 2010
[12]

Gene regulatory landscape dissected by single-cell four-omics sequencing

Yujie Chen et al. “Gene regulatory landscape dissected by single-cell four-omics sequencing”. In:Nature(2026), pp. 1–10

2026
[13]

Evolutionary constraints on the complexity of genetic regulatory networks allow predictions of the total number of genetic interactions

Adrian I. Campos and Julio A. Freyre-González. “Evolutionary constraints on the complexity of genetic regulatory networks allow predictions of the total number of genetic interactions”. en. In:Sci Rep9.1 (Mar. 2019). Publisher: Nature Publishing Group, p. 3618.ISSN: 2045-2322.DOI: 10.1038/s41598-019-39866-z . URL:https://www.nature.com/articles/s41598-019-39866-z

work page doi:10.1038/s41598-019-39866-z 2019
[14]

A meta-analysis of Boolean network models reveals design principles of gene regulatory networks

Claus Kadelka et al. “A meta-analysis of Boolean network models reveals design principles of gene regulatory networks”. en. In:Sci. Adv.10.2 (Jan. 2024), eadj0822.ISSN: 2375-2548.DOI: 10.1126/sciadv.adj0822 . URL:https://www.science.org/doi/10.1126/sciadv.adj0822

work page doi:10.1126/sciadv.adj0822 2024
[15]

Architecture of the human regulatory network derived from ENCODE data

Mark B. Gerstein et al. “Architecture of the human regulatory network derived from ENCODE data”. en. In:Nature489.7414 (Sept. 2012). Publisher: Nature Publishing Group, pp. 91–100.ISSN: 1476-4687.DOI: 10.1038/nature11245.URL:https://www.nature.com/articles/nature11245

work page doi:10.1038/nature11245.url:https://www.nature.com/articles/nature11245 2012
[16]

Hopfield

J J Hopfield. “Neural networks and physical systems with emergent collective computational abilities.” In: Proceedings of the National Academy of Sciences79.8 (Apr. 1982). Publisher: Proceedings of the National Academy of Sciences, pp. 2554–2558.DOI: 10.1073/pnas.79.8.2554 .URL: https://www.pnas.org/ doi/10.1073/pnas.79.8.2554

work page doi:10.1073/pnas.79.8.2554 1982
[17]

Intrinsic noise in gene regulatory networks

Mukund Thattai and Alexander van Oudenaarden. “Intrinsic noise in gene regulatory networks”. In:Proceedings of the National Academy of Sciences of the United States of America98.15 (July 2001), pp. 8614–8619.ISSN: 0027-8424.DOI: 10 . 1073 / pnas . 151588598.URL: https : / / pmc . ncbi . nlm . nih . gov / articles / PMC37484/

2001
[18]

Stochastic Gene Expression in a Single Cell

Michael B. Elowitz et al. “Stochastic Gene Expression in a Single Cell”. en. In:Science297.5584 (Aug. 2002). Publisher: American Association for the Advancement of Science (AAAS), pp. 1183–1186.ISSN: 0036-8075, 1095-9203.DOI: 10.1126/science.1070919.URL: https://www.science.org/doi/10.1126/science. 1070919

work page doi:10.1126/science.1070919.url: 2002
[19]

Role of noise and parametric variation in the dynamics of gene regulatory circuits

Vivek Kohar and Mingyang Lu. “Role of noise and parametric variation in the dynamics of gene regulatory circuits”. en. In:npj Systems Biology and Applications4.1 (Nov. 2018). Publisher: Nature Publishing Group, p. 40.ISSN: 2056-7189.DOI: 10.1038/s41540-018-0076-x .URL: https://www.nature.com/articles/ s41540-018-0076-x

work page doi:10.1038/s41540-018-0076-x 2018
[20]

ISSN: 2692-8205 Pages: 2025.07.22.666238 Section: New Results

James Holehouse.Do Distinct Subpopulations Signify Modes of Behavior in a Noisy Single Cell?en. ISSN: 2692-8205 Pages: 2025.07.22.666238 Section: New Results. July 2025.DOI: 10.1101/2025.07.22.666238. URL:https://www.biorxiv.org/content/10.1101/2025.07.22.666238v1

work page doi:10.1101/2025.07.22.666238 2025
[21]

Extrinsic Noise and Heavy-Tailed Laws in Gene Expression

Lucy Ham, Rowan D. Brackston, and Michael P. H. Stumpf. “Extrinsic Noise and Heavy-Tailed Laws in Gene Expression”. en. In:Phys. Rev. Lett.124.10 (Mar. 2020), p. 108101.ISSN: 0031-9007, 1079-7114.DOI: 10.1103/ PhysRevLett.124.108101.URL:https://link.aps.org/doi/10.1103/PhysRevLett.124.108101

work page doi:10.1103/physrevlett.124.108101 2020
[22]

Gene Expression Plasticity Is Associated with Regulatory Complexity but Not with Specific Network Motifs

Apolline J R Petit, Anne Genissel, and Arnaud Le Rouzic. “Gene Expression Plasticity Is Associated with Regulatory Complexity but Not with Specific Network Motifs”. en. In:bioRxiv(June 2025).DOI: https : //doi.org/10.1101/2024.03.11.584403. 15

work page doi:10.1101/2024.03.11.584403 2025
[23]

Network motifs: theory and experimental approaches

Uri Alon. “Network motifs: theory and experimental approaches”. en. In:Nature Reviews Genetics8.6 (June 2007), pp. 450–461.ISSN: 1471-0056, 1471-0064.DOI: 10.1038/nrg2102 .URL: https://www.nature. com/articles/nrg2102

work page doi:10.1038/nrg2102 2007
[24]

Gene regulatory networks and the role of robustness and stochasticity in the control of gene expression

Lesley T. MacNeil and Albertha J. M. Walhout. “Gene regulatory networks and the role of robustness and stochasticity in the control of gene expression”. en. In:Genome Research21.5 (May 2011). Company: Cold Spring Harbor Laboratory Press Distributor: Cold Spring Harbor Laboratory Press Institution: Cold Spring Harbor Laboratory Press Label: Cold Spring Har...

work page doi:10.1101/gr.097378.109.url: 2011
[25]

Canalization reduces the nonlinearity of regulation in biological net- works

Claus Kadelka and David Murrugarra. “Canalization reduces the nonlinearity of regulation in biological net- works”. en. In:npj Syst Biol Appl10.1 (June 2024). Publisher: Nature Publishing Group, p. 67.ISSN: 2056-7189. DOI: 10.1038/s41540-024-00392-y .URL: https://www.nature.com/articles/s41540-024-00392- y

work page doi:10.1038/s41540-024-00392-y 2024
[26]

arXiv:1111.2759 [q-bio]

Lori Layne, Elena Dimitrova, and Matthew Macauley.Nested canalyzing depth and network stability. arXiv:1111.2759 [q-bio]. Nov. 2011.DOI: 10 . 48550 / arXiv . 1111 . 2759.URL: http : / / arxiv . org / abs/1111.2759

work page arXiv 2011
[27]

Adaptation to an extraordinary environment by evolution of phenotypic plasticity and genetic assimilation

R. Lande. “Adaptation to an extraordinary environment by evolution of phenotypic plasticity and genetic assimilation”. In:Journal of Evolutionary Biology22.7 (July 2009), pp. 1435–1446.ISSN: 1010-061X.DOI: 10. 1111/j.1420-9101.2009.01754.x.URL:https://doi.org/10.1111/j.1420-9101.2009.01754.x

work page doi:10.1111/j.1420-9101.2009.01754.x 2009
[28]

EVOLUTION AND EXTINCTION IN A CHANGING ENVIRONMENT: A QUANTITATIVE-GENETIC ANALYSIS

Reinhard Bürger and Michael Lynch. “EVOLUTION AND EXTINCTION IN A CHANGING ENVIRONMENT: A QUANTITATIVE-GENETIC ANALYSIS”. In:Evolution49.1 (Feb. 1995), pp. 151–163.ISSN: 0014-3820. DOI: 10.1111/j.1558-5646.1995.tb05967.x .URL: https://doi.org/10.1111/j.1558-5646.1995. tb05967.x

work page doi:10.1111/j.1558-5646.1995.tb05967.x 1995
[29]

Evolution takes multiple paths to evolvability when facing environmental change

Bhaskar Kumawat et al. “Evolution takes multiple paths to evolvability when facing environmental change”. In: Proceedings of the National Academy of Sciences122.1 (Jan. 2025). Publisher: Proceedings of the National Academy of Sciences, e2413930121.DOI: 10.1073/pnas.2413930121 .URL: https://www.pnas.org/ doi/abs/10.1073/pnas.2413930121

work page doi:10.1073/pnas.2413930121 2025
[30]

Evolving Evolvability in the Context of Environmental Change: A Gene Regulatory Network (GRN) Approach

Yifei Wang, Joanna Bryson, and Stephen Matthews. “Evolving Evolvability in the Context of Environmental Change: A Gene Regulatory Network (GRN) Approach”. en. In: MIT Press, July 2014, pp. 47–53.DOI: 10.1162/978-0-262-32621-6-ch010 .URL: https://dx.doi.org/10.1162/978-0-262-32621-6- ch010

work page doi:10.1162/978-0-262-32621-6-ch010 2014
[31]

Daniel M Weinreich et al.The population genetics of biological noise. en. Jan. 2025.DOI: 10.1101/2025.01. 11.632402.URL:http://biorxiv.org/lookup/doi/10.1101/2025.01.11.632402

work page doi:10.1101/2025.01 2025
[32]

Evolution of gene networks by gene duplications: a mathematical model and its implications on genome organization

A Wagner. “Evolution of gene networks by gene duplications: a mathematical model and its implications on genome organization.” In:Proceedings of the National Academy of Sciences91.10 (May 1994). Publisher: Proceedings of the National Academy of Sciences, pp. 4387–4391.DOI: 10.1073/pnas.91.10.4387.URL: https://www.pnas.org/doi/10.1073/pnas.91.10.4387

work page doi:10.1073/pnas.91.10.4387.url: 1994
[33]

DOES EVOLUTIONARY PLASTICITY EVOLVE?

Andreas Wagner. “DOES EVOLUTIONARY PLASTICITY EVOLVE?” en. In:Evolution50.3 (June 1996), pp. 1008–1023.ISSN: 00143820.DOI: 10.1111/j.1558-5646.1996.tb02342.x.URL: https://academic. oup.com/evolut/article/50/3/1008/6870974

work page doi:10.1111/j.1558-5646.1996.tb02342.x.url: 1996
[34]

Extrinsic noise and heavy-tailed laws in gene expression

Lucy Ham, Rowan D Brackston, and Michael PH Stumpf. “Extrinsic noise and heavy-tailed laws in gene expression”. In:Physical review letters124.10 (2020), p. 108101

2020
[35]

Quantifying and correcting bias in transcriptional parameter inference from single-cell data

Ramon Grima and Pierre-Marie Esmenjaud. “Quantifying and correcting bias in transcriptional parameter inference from single-cell data”. In:Biophysical Journal123.1 (2024), pp. 4–30

2024
[36]

Quantifying broken detailed balance in transcription

James Holehouse. “Quantifying broken detailed balance in transcription”. In:npj Complexity3.1 (2026), p. 10

2026
[37]

Gravitationally induced decoherence vs space-time diffusion: testing the quantum nature of gravity.Nature Commun., 14(1):7910, 2023

Yipei Guo and Ariel Amir. “Exploring the effect of network topology, mRNA and protein dynamics on gene regulatory network stability”. en. In:Nat Commun12.1 (Jan. 2021). Publisher: Nature Publishing Group, p. 130. ISSN: 2041-1723.DOI: 10.1038/s41467- 020- 20472- x.URL: https://www.nature.com/articles/ s41467-020-20472-x

work page doi:10.1038/s41467- 2021
[38]

Motifs emerge from function in model gene regulatory networks

Z. Burda et al. “Motifs emerge from function in model gene regulatory networks”. en. In:Proc. Natl. Acad. Sci. U.S.A.108.42 (Oct. 2011), pp. 17263–17268.ISSN: 0027-8424, 1091-6490.DOI: 10.1073/pnas.1109435108. URL:https://pnas.org/doi/full/10.1073/pnas.1109435108

work page doi:10.1073/pnas.1109435108 2011
[39]

Nature Reviews Genetics , volume=

Albert-László Barabási and Zoltán N. Oltvai. “Network biology: understanding the cell’s functional organization”. en. In:Nat Rev Genet5.2 (Feb. 2004), pp. 101–113.ISSN: 1471-0056, 1471-0064.DOI: 10.1038/nrg1272. URL:https://www.nature.com/articles/nrg1272. 16

work page doi:10.1038/nrg1272 2004
[40]

Spontaneous evolution of modularity and network motifs

Nadav Kashtan and Uri Alon. “Spontaneous evolution of modularity and network motifs”. In:Proceedings of the National Academy of Sciences102.39 (Sept. 2005), pp. 13773–13778.DOI: 10.1073/pnas.0503610102.URL: https://www.pnas.org/doi/10.1073/pnas.0503610102

work page doi:10.1073/pnas.0503610102.url: 2005
[41]

Arnaud Le Rouzic and Sylvain Pouzet.Gene network topology drives the mutational landscape of gene expression. en. Nov. 2024.DOI: 10.1101/2024.11.28.625874.URL: http://biorxiv.org/lookup/doi/10.1101/ 2024.11.28.625874

work page doi:10.1101/2024.11.28.625874.url: 2024
[42]

Waddington’s canalization revisited: Developmental stability and evolution

Mark L. Siegal and Aviv Bergman. “Waddington’s canalization revisited: Developmental stability and evolution”. In:Proceedings of the National Academy of Sciences99.16 (Aug. 2002). Publisher: Proceedings of the National Academy of Sciences, pp. 10528–10532.DOI: 10.1073/pnas.102303999.URL: https://www.pnas.org/ doi/full/10.1073/pnas.102303999

work page doi:10.1073/pnas.102303999.url: 2002
[43]

Intrinsic and extrinsic contributions to stochasticity in gene expression

Peter S Swain, Michael B Elowitz, and Eric D Siggia. “Intrinsic and extrinsic contributions to stochasticity in gene expression”. In:Proceedings of the National Academy of Sciences99.20 (2002), pp. 12795–12800

2002
[44]

Phenotypic consequences of promoter-mediated transcriptional noise

William J Blake et al. “Phenotypic consequences of promoter-mediated transcriptional noise”. In:Molecular cell 24.6 (2006), pp. 853–865

2006
[45]

Quantifying E. coli proteome and transcriptome with single-molecule sensitivity in single cells

Yuichi Taniguchi et al. “Quantifying E. coli proteome and transcriptome with single-molecule sensitivity in single cells”. In:science329.5991 (2010), pp. 533–538

2010
[46]

Control of transcript variability in single mammalian cells

Nico Battich, Thomas Stoeger, and Lucas Pelkmans. “Control of transcript variability in single mammalian cells”. In:Cell163.7 (2015), pp. 1596–1610

2015
[47]

A regulatory hierarchy controls the dynamic transcriptional response to extreme oxidative stress in archaea

Peter D Tonner et al. “A regulatory hierarchy controls the dynamic transcriptional response to extreme oxidative stress in archaea”. In:PLoS Genetics11.1 (2015), e1004912

2015
[48]

Stochastic Modeling of Autoregulatory Genetic Feedback Loops: A Review and Comparative Study

James Holehouse, Zhixing Cao, and Ramon Grima. “Stochastic Modeling of Autoregulatory Genetic Feedback Loops: A Review and Comparative Study”. English. In:Biophysical Journal118.7 (Apr. 2020). Publisher: Elsevier, pp. 1517–1525.ISSN: 0006-3495, 1542-0086.DOI: 10.1016/j.bpj.2020.02.016 .URL: https: //www.cell.com/biophysj/abstract/S0006-3495(20)30165-X

work page doi:10.1016/j.bpj.2020.02.016 2020
[49]

Noise in Gene Regulatory Networks

Ioannis Lestas et al. “Noise in Gene Regulatory Networks”. In:IEEE Transactions on Automatic Control 53.Special Issue (Jan. 2008), pp. 189–200.ISSN: 1558-2523.DOI: 10 . 1109 / TAC . 2007 . 911347.URL: https://ieeexplore.ieee.org/abstract/document/4439816

work page arXiv 2008
[50]

Network Motifs: Simple Building Blocks of Complex Networks

R Milo et al. “Network Motifs: Simple Building Blocks of Complex Networks”. en. In: (2002)

2002
[51]

Biological network motif detection: principles and practice

E. Wong et al. “Biological network motif detection: principles and practice”. en. In:Briefings in Bioinformatics 13.2 (Mar. 2012), pp. 202–215.ISSN: 1467-5463, 1477-4054.DOI: 10 . 1093 / bib / bbr033.URL: https : //academic.oup.com/bib/article-lookup/doi/10.1093/bib/bbr033

work page doi:10.1093/bib/bbr033 2012
[52]

Should evolutionary geneticists worry about higher-order epistasis?

Daniel M Weinreich et al. “Should evolutionary geneticists worry about higher-order epistasis?” In:Current Opinion in Genetics & Development. Genetics of system biology 23.6 (Dec. 2013), pp. 700–707.ISSN: 0959- 437X

2013
[53]

Adaptive Landscape by Environment Interactions Dictate Evolutionary Dynamics in Models of Drug Resistance

C. Brandon Ogbunugafor et al. “Adaptive Landscape by Environment Interactions Dictate Evolutionary Dynamics in Models of Drug Resistance”. en. In:PLOS Computational Biology12.1 (Jan. 2016), e1004710.ISSN: 1553- 7358

2016
[54]

Epistasis — the essential role of gene interactions in the structure and evolution of genetic systems

Patrick C. Phillips. “Epistasis — the essential role of gene interactions in the structure and evolution of genetic systems”. en. In:Nature Reviews Genetics9.11 (Nov. 2008), pp. 855–867.ISSN: 1471-0056, 1471-0064.DOI: 10.1038/nrg2452.URL:https://www.nature.com/articles/nrg2452

work page doi:10.1038/nrg2452.url:https://www.nature.com/articles/nrg2452 2008
[55]

A new frontier for Hopfield networks

Dmitry Krotov. “A new frontier for Hopfield networks”. In:Nature Reviews Physics5.7 (2023), pp. 366–367

2023
[56]

Hopfield networks is all you need.arXiv preprint arXiv:2008.02217, 2020

Hubert Ramsauer et al. “Hopfield networks is all you need”. In:arXiv preprint arXiv:2008.02217(2020)

work page arXiv 2008
[57]

Noise characteristics of feed forward loops

Bhaswar Ghosh, Rajesh Karmakar, and Indrani Bose. “Noise characteristics of feed forward loops”. en. In: Physical Biology2.1 (Mar. 2005), p. 36.ISSN: 1478-3975.DOI: 10 . 1088 / 1478 - 3967 / 2 / 1 / 005.URL: https://doi.org/10.1088/1478-3967/2/1/005

work page doi:10.1088/1478-3967/2/1/005 2005
[58]

Buffered Qualitative Stability explains the robustness and evolvability of transcriptional networks

Luca Albergante, J Julian Blow, and Timothy J Newman. “Buffered Qualitative Stability explains the robustness and evolvability of transcriptional networks”. In:eLife3 (Sept. 2014). Ed. by Detlef Weigel, e02863.ISSN: 2050-084X.DOI:10.7554/eLife.02863.URL:https://doi.org/10.7554/eLife.02863

work page doi:10.7554/elife.02863.url:https://doi.org/10.7554/elife.02863 2014
[59]

Are network motifs the spandrels of cellular complexity?

R Sole and S Valverde. “Are network motifs the spandrels of cellular complexity?” en. In:Trends in Ecology & Evolution21.8 (Aug. 2006), pp. 419–422.ISSN: 01695347.DOI: 10.1016/j.tree.2006.05.013 .URL: https://linkinghub.elsevier.com/retrieve/pii/S0169534706001674

work page doi:10.1016/j.tree.2006.05.013 2006
[60]

Catch Me if You Can: Adaptation from Standing Genetic Variation to a Moving Phenotypic Optimum

Sebastian Matuszewski, Joachim Hermisson, and Michael Kopp. “Catch Me if You Can: Adaptation from Standing Genetic Variation to a Moving Phenotypic Optimum”. In:Genetics200.4 (Aug. 2015), pp. 1255–1274. ISSN: 0016-6731.DOI: 10 . 1534 / genetics . 115 . 178574.URL: https : / / pmc . ncbi . nlm . nih . gov / articles/PMC4574244/. 17

2015
[61]

Errors during Gene Expression: Single-Cell Heterogeneity, Stress Resistance, and Microbe-Host Interactions

Christopher R. Evans et al. “Errors during Gene Expression: Single-Cell Heterogeneity, Stress Resistance, and Microbe-Host Interactions”. In:mBio9.4 (July 2018), e01018–18.ISSN: 2150-7511.DOI: 10.1128/mBio. 01018-18.URL:https://pmc.ncbi.nlm.nih.gov/articles/PMC6030554/

work page doi:10.1128/mbio 2018
[62]

With a Little Help from Physics

Anil Ananthaswamy. “With a Little Help from Physics”. en. In:Why Machines Learn. New York: DUTTON, 2024. 18 A Parameters and Code Availability Table 1: Standard parameters used in the simulation. Symbol Description Value GNumber of generations per run 500 MMax steps before declaring a state unstable 100 sSelection pressure 10 unstable_fitness Fitness for ...

2024