{"id":"1d0c713d-7228-4e0e-a1ba-1e70a98f6bbf","arxiv_id":"2411.13630","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"For a gene controlled by competing activator and repressor, this paper derives analytic noise formulas, fits promoter rates to published dose-response data, and shows reinitiation shifts noise between mRNA and protein.","lead":"This paper builds a stochastic model of a gene where activator and repressor proteins compete for the same promoter site, and derives equations for average mRNA and protein levels and for expression noise with and without transcriptional reinitiation. It fits promoter rates to a classic dose-response experiment and predicts that the competitive circuit is noisier above the Poisson baseline but can be pushed further below it than the non-competitive circuit.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The Fano factor formula for the with-reinitiation competitive model (Eq. 27) contains undefined symbols h2, h6, h8, so the paper's central noise-ordering claim cannot be verified from the printed derivation.","rationale":"The reader's weakest assumption—parameter transferability—is a valid external-validity concern, and the paper's reliance on Blake rates without noise data would indeed make the quantitative noise predictions conditional. But the more immediate, internally checkable problem is that the central analytical objects (Eqs. 27 and 31) are not self-contained. A formula with undefined symbols cannot be independently reproduced, so the noise-ordering claim—the part of the abstract and Section 6 that is presented as the paper's main new result—is not verifiable from the text. This is fixable, and the underlying CME calculation may well be correct; hence the verdict remains CONDITIONAL rather than REJECT. I partially agree with the reader: the reader noted the undefined symbols in the rationale but placed parameter transferability as the weakest assumption; I regard the incomplete formula as the single most load-bearing issue.","tokens_in":19626,"tokens_out":19141,"duration_ms":191483,"concrete_test":"Independently derive F F_m^CWR from the generating-function equations in Appendix A for the four-state scheme (Gr, Gn, Ga, Gc), matching the moments used to obtain Eq. 27 and identifying the intended h2, h6, h8. Then recompute Figs. 10c and 11c with the same Blake-parameter values; if the reconstructed Fano factor differs from Eq. 27 in any non-negligible term, the noise-ordering conclusions must be recalculated and the current figures may not support the abstract claims.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's headline noise-ordering claims—competitive Fano factor above non-competitive in the super-Poissonian regime, and reinitiation lowering mRNA Fano while raising protein Fano—rest on the analytical expressions in Eqs. 27–31. Eq. 27 for the with-reinitiation mRNA Fano factor is not a closed expression: h2, h6, and h8 appear in both numerator and denominator but are never defined in the main text, Appendix A, the glossary, or the parameter list. The surrounding l_i and r_i variables are defined, but no h_i are. Eq. 31 similarly defines X through a chain (r6–r10) that contains an internal typo (r7 is assigned twice and used in r10), so the without-reinitiation protein Fano factor is also not independently checkable. These are not mere style issues: the central quantitative predictions are computed from equations the reader cannot evaluate. Without the missing definitions or an alternate derivation, the noise-ordering claim is not established as printed.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops an analytical stochastic model of a three-state competitive activator-repressor promoter, with and without transcriptional reinitiation. The authors derive mean mRNA/protein levels and Fano factors via a generating-function approach, estimate dox-dependent rate parameters by fitting the dose-response data of Rossi et al. [36], compare the competitive circuit with a non-competitive architecture, and report noise-ordering results: higher mRNA Fano factor in the super-Poissonian regime for the competitive circuit, lower mRNA but higher protein Fano factor with reinitiation than without, and the possibility of reducing noise below the Poissonian level by lowering k4.","tokens_in":19948,"tokens_out":3576,"duration_ms":41389,"significance":"If the central claims hold, the paper would provide a useful analytical framework and a parameter set for a circuit that has previously been studied mainly through Hill-function fits or simulations. The generating-function formalism is standard, and the authors supplement the analysis with Gillespie simulations and a sensitivity/chi-square check, which are appropriate tools. The main significance, however, rests on the noise-ordering statements in Secs. 5.2 and 6; as printed, those statements cannot be verified because key analytical expressions contain undefined symbols and because the noise predictions are computed with rate constants transferred from a different experimental system. The claimed theoretical derivation of the dox power 3.4 is also circular as presented, since the exponents 1.6 and 1.8 are read from the very data being explained.","major_comments":[{"comment":"The with-reinitiation mRNA Fano factor is not a closed expression: h2, h6, and h8 appear in both numerator and denominator but are never defined in the main text, Appendix A, the glossary, or the parameter list. Since the paper's central noise-ordering claims are computed from this equation, the reader cannot independently evaluate or verify them. Please define these quantities or provide an alternative derivation, a symbolic algorithm, or machine-readable code that generates the expression.","section":"§5.2, Eq. (27)"},{"comment":"The protein Fano factor expressions are also not independently checkable as printed. Eq. (31) defines X through a chain of r-variables in which r7 is assigned twice, and the derivation of the chain is not shown anywhere. Even if the duplicated assignment is a harmless typo, the lack of a derivation means the without-reinitiation protein Fano factor, which underpins the reinitiation noise-ordering claims in Figs. 10–11, cannot be verified from the manuscript.","section":"§5.2, Eqs. (28)–(31)"},{"comment":"The statement that the dox power 3.4 is 'theoretically obtained' is not supported. The exponents 1.6 and 1.8 are explicitly chosen in Sec. 3.2 and Sec. 4 'following the Hill coefficient values found in [36]', so the product S^1.6 S^1.8 = S^3.4 is a product of two experimentally fitted exponents, not an independent prediction. The agreement with the observed Hill coefficient 3.2 should be reframed as a consistency check of the proposed multiplicative composition rule, and ideally tested against independent data or against a model in which the dox powers are free parameters.","section":"§5.1"},{"comment":"The noise-ordering conclusions are numerical results obtained with rate constants transferred from Blake et al.'s yeast GAL/aTc system, while the dox-dependent parameters and Zth are taken from fits to Rossi et al.'s dox data. The manuscript provides no evidence that these two parameter sets describe the same promoter chemistry. The claimed ordering of competitive versus non-competitive Fano factors and the reinitiation anomalies may therefore be artifacts of the chosen parameter values rather than architectural properties. Please test robustness by varying the transferred rates over a plausible range, or by using the Rossi-derived rates in the stochastic analysis.","section":"§6.1 and Figs. 10–11"},{"comment":"The statement that a 'most probable set of parameter values' is found is stronger than the evidence supports. The manuscript says Zth can be set to any value between 70% and 99% to obtain a best fit, so the parameter set is not unique. The sensitivity and MSE analysis in Appendix B is welcome, but the paper should clearly state the resulting confidence intervals for the fitted parameters and explicitly acknowledge the degeneracy implied by the free threshold.","section":"§5.1"}],"minor_comments":[{"comment":"The glossary defines the Fano factor as 'variance of protein / mean mRNA', which is dimensionally and conceptually inconsistent with the rest of the paper, where the mRNA Fano factor is variance of mRNA divided by mean mRNA. Please correct the definition and remove the duplicated wording.","section":"Glossary"},{"comment":"In the abstract, 'there exits some anomalous characteristic features' should be 'there exist some anomalous characteristic features'.","section":"Abstract"},{"comment":"The hypergeometric expression for the activator-only promoter activity is very hard to parse because of the compressed notation, including 'J kd 1' in the denominator. Please restructure the equation with clear definitions of all arguments.","section":"Eq. (8)"},{"comment":"In the master equation, the mRNA degradation term reads 'P (n1, n2, n3n4 + 1, n5, t)', which appears to be missing a comma before n4. Please fix this typo.","section":"§5.2, Eq. (25)"},{"comment":"Reference [12] appears to be a geophysics paper on post-glacial rebound and seems unrelated to stochastic gene expression. Please verify and correct the citation.","section":"References"},{"comment":"The declaration heading contains a typo: 'Competiting Interests' should be 'Competing Interests'.","section":"Declarations"}],"recommendation":"major_revision","confidential_remarks":"The manuscript addresses an interesting and potentially useful system, and the mean-level algebra in the two-state reductions appears consistent. My concern is that the main quantitative claims—the noise-ordering between competitive and non-competitive architectures and the sign of the reinitiation effect—cannot be checked from the printed equations and depend on a cross-system parameter transfer that is not defended. These issues are fixable in principle, but they are load-bearing rather than cosmetic. I would also encourage the authors to tone down the 'theoretical prediction' language for the dox power 3.4, since it is obtained from fitted exponents."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick read for you on Das & Biswas, arXiv:2411.13630. The genuinely new piece is a fitted parameter set for the Rossi et al. competitive activator-repressor promoter and analytic Fano-factor expressions for the three-state model with transcriptional reinitiation; the noise comparison between competitive and non-competitive architectures (with and without reinitiation) is also new. The mean mRNA expression checks out algebraically, the parameter estimation is done with care (relative error, MSE, sensitivity), and the Gillespie simulations give the authors' fits some independent support.\n\nThe soft spots are real but not fatal in principle. First, the load-bearing analytical expressions for the with-reinitiation Fano factors are not self-contained: Eq. 27 contains undefined symbols h2, h6, h8 in both numerator and denominator, and Eq. 31's X is built from a chain with an internal typo (r7 assigned twice). As printed, the central noise-ordering claims—competitive noise above non-competitive in the super-Poissonian regime, reinitiation lowering mRNA noise while raising protein noise—cannot be independently checked. That is a reproducibility problem, not a style quibble. Second, the 'theoretical' Hill power of 3.4 is obtained by multiplying the dox exponents 1.6 and 1.8 that were themselves read from the same Rossi data; calling it theoretical overstates what is really a composition of fitted forms. Third, the noise predictions use rate constants borrowed from Blake et al. (yeast GAL/aTc) and are not validated against measured noise, so the architecture-level ordering could in principle be an artifact of parameter choice.\n\nNone of this sinks the paper. The CME approach is standard, the mean expressions are consistent, and the fitted parameter set is a useful contribution that has been missing from the literature. The undefined symbols and the typo are fixable in revision—they need a supplementary derivation or a corrected appendix.\n\nWho is this for? People working on stochastic gene-expression models and synthetic gene circuits who want a concrete parameter set for a competitive promoter and a starting point for noise calculations. It deserves a serious referee: the topic is important enough within the subfield, and the problems are fixable. I would not desk-reject it, but I would insist the authors supply complete definitions for the Fano-factor equations and fix the typo before it is published. My verdict tracks the reader's 'conditional'.","headline":"Useful fitted parameter set and Fano-factor formulas for a competitive activator-repressor promoter, but the central noise formulas are not checkable as printed due to undefined symbols, so the paper needs a corrected revision before it can be fully evaluated.","tokens_in":20440,"tokens_out":2420,"would_cite":false,"duration_ms":25271,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["87.10.Mn"],"model":"deepseek-v4-flash","headline":"The paper derives the full kinetics of a gene whose activator and repressor compete for one promoter site, reproduces the observed steep dox dose-response (power 3.4 vs 3.2), and predicts the noise cost of competitive regulation.","keywords":["stochastic gene expression","competitive transcription factor binding","activator-repressor system","Fano factor","transcriptional reinitiation","dose-response curve","Hill coefficient","parameter estimation"],"falsifier":"A direct test is a single-cell measurement of the mRNA Fano factor in the dox-controlled competitive circuit of [36] alongside a matched non-competitive circuit: the noise-ordering claim fails if the competitive circuit is not noisier in the super-Poissonian regime. A second test is the dose-response exponent itself: if the competitive dose-response saturates with a dox power clearly different from $3.4$, the rate-factorization mechanism $k_{\\rm ON}=k_1 k_a$ is wrong.","tokens_in":19388,"feed_emoji":"🧬","tokens_out":18456,"duration_ms":154044,"temperature":0.7,"pith_summary":"This paper tries to establish a complete analytical theory of a gene whose activator and repressor bind the promoter competitively — a circuit measured experimentally in [36] but never given a full kinetic parameter set. It claims the effective promoter ON and OFF rates factor into products and sums of inducer-dependent single-molecule rates, so the inducer's power-law exponents add, yielding a theoretical dox power of $3.4$ that matches the observed Hill coefficient $3.2$ [36]. Using the derived rates, it computes exact Fano factors (variance-to-mean ratios) for mRNA and protein and claims the competitive architecture is noisier than the non-competitive one in the super-Poissonian regime, while transcriptional reinitiation lowers mRNA-level noise and raises protein-level noise. If correct, the paper supplies the missing parameter set, explains the steep all-or-none response as an emergent property of competitive binding, and makes noise predictions that synthetic designs of such switches could test.","feed_headline":"Competitive gene control steepens the switch: 1.6 + 1.8 = 3.4","feed_subtitle":"Theory recreates the steep all-or-none switch and spells out the noise cost of molecular competition.","key_machinery":"The load-bearing object is the three-state gene model $G_r \\to G_n \\to G_a$, in which repressor-related transition rates $k_1$, $k_2$ compete with activator-related rates $k_a$, $k_d$, extended by an initiation-complex state $G_c$ for transcriptional reinitiation ($k_3$, $k_4$). The argument is carried by the algebraic reduction $k_{\\rm ON}=k_1 k_a$, $k_{\\rm OFF}=k_d(k_1+k_2)$, which maps the three-state scheme onto an equivalent two-state promoter and makes inducer powers additive, and by a moment-generating-function calculation that yields the exact Fano-factor formulas (Eqs. 27-31).","core_discovery":"The paper's central claim is that a three-state promoter in which activator and repressor bind mutually exclusively can be collapsed to an effective two-state circuit whose ON and OFF rates factor as $k_{\\rm ON}=k_1 k_a$ and $k_{\\rm OFF}=k_d(k_1+k_2)$. Since each component rate is a power law in the inducer dox, the effective exponents add: the activator branch carries $S^{1.6}$, the repressor branch $S^{1.8}$, so the competitive dose-response should scale as $S^{3.4}$, close to the experimentally observed Hill coefficient of $3.2$ [36]. From the same framework the paper derives closed-form Fano factors (variance-to-mean ratios) for mRNA and protein, with and without transcriptional reinitiation. It then claims the competitive circuit is noisier than the non-competitive one in the super-Poissonian regime ($F>1$), that reinitiation lowers mRNA-level noise while raising protein-level noise, and that tightening RNAP-II retention ($k_4$) pushes noise below the Poissonian level more strongly in the competitive circuit than in the non-competitive one.","pith_inferences":["Editorial extension: the factorization rule $k_{\\rm ON}=k_1 k_a$ is generic, so any pair of inducers with power-law binding rates should show a summed effective exponent in a competitive circuit; this is testable with inducer pairs other than dox.","Editorial extension: the predicted competitive-versus-non-competitive noise ordering could be checked against existing single-molecule mRNA data from comparable synthetic TetR/GAL circuits, without building new experiments.","Editorial extension: the inducer values where with-reinitiation and without-reinitiation curves cross in mean and Fano factor are a measurable signature of reinitiation strength and could calibrate $k_3$ and $k_4$ in live cells.","Editorial extension: the finding that competitive circuits reach lower sub-Poissonian noise suggests a design heuristic — mutually exclusive TF binding plus tight RNAP-II retention — for noise-suppressed synthetic gene circuits."],"forward_implications":["The fitted rate set ($k_1$, $k_2$, $k_a$, $k_d$, $J_m$, $Z_{\\rm th}$) gives experimenters and simulators the first concrete parameters for the dox-controlled competitive circuit, replacing an earlier absence of kinetic rates.","Because mean expression levels coincide under identical rate constants, the higher super-Poissonian noise of the competitive circuit is an architectural property and not an artifact of different expression levels.","Reinitiation's opposed effects on mRNA and protein Fano factors mean the two noise levels cannot be tuned independently through reinitiation alone.","Reducing $k_4$, interpreted as tighter RNAP-II binding to the promoter, is predicted to drive mRNA noise below the Poissonian level in both circuits, with the competitive circuit reaching lower sub-Poissonian values.","The theoretical dox power $3.4$ matching the measured $3.2$ means the steep switch response does not require a Hill-function assumption: it emerges from multiplicative rate composition."],"supporting_citations":[{"why":"Supplies the experimental dox dose-response data and the observed Hill coefficients 1.6, 1.8, and 3.2 that the paper's parameter estimation must fit.","marker":"[36]"},{"why":"Supplies the GAL/aTc-dependent rate-constant forms and the interpretation of k4 as RNAP-II binding tightness used in the stochastic noise analysis.","marker":"[3]"},{"why":"Provides the three-state activator-repressor model of graded-to-binary response whose probability distribution approach the paper extends to full kinetics and noise.","marker":"[16]"},{"why":"The earlier stochastic Markov chain model that left the Hill-coefficient mismatch and the kinetics unexplained, motivating the present theory.","marker":"[37]"},{"why":"The non-competitive activator-repressor architecture that serves as the comparison baseline for mean expression and Fano factors.","marker":"[41]"},{"why":"The stochastic simulation algorithm used to verify the analytical dose-response curves against the experimental data.","marker":"[34]"},{"why":"The reinitiation noise-control theory whose treatment of transcriptional reinitiation the present paper incorporates and extends.","marker":"[32]"}],"fun_headline_variants":["Competitive gene switch steepens: 1.6 + 1.8 = 3.4","Competitive TF binding: noise reduction and steeper gene switch","Two-state reduction explains competitive gene circuit noise","Activator and repressor exponents add: gene switch Hill ~3.4"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that kinetic rate constants and functional forms measured in one system — the yeast GAL/aTc rates of [3] — transfer faithfully to a different circuit, the dox-driven competitive promoter of [36]; if the true competitive rates follow different chemistry or concentration dependence, the predicted noise ordering between competitive and non-competitive circuits could be a parameter artifact rather than a property of the architecture.","fun_headline_variants_meta":{"raw":{"variants":["Competitive gene switch steepens: 1.6 + 1.8 = 3.4","Competitive TF binding: noise reduction and steeper gene switch","Two-state reduction explains competitive gene circuit noise","Activator and repressor exponents add: gene switch Hill ~3.4"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001417,"raw_usage":{"total_tokens":5758,"prompt_tokens":1019,"completion_tokens":4739,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":635,"completion_tokens_details":{"reasoning_tokens":4660}},"tokens_in":635,"tokens_out":4739,"duration_ms":34703,"temperature":1.0,"reasoning_tokens":4660,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T16:19:40.538779+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct test is a single-cell measurement of the mRNA Fano factor in the dox-controlled competitive circuit of [36] alongside a matched non-competitive circuit: the noise-ordering claim fails if the competitive circuit is not noisier in the super-Poissonian regime. A second test is the dose-response exponent itself: if the competitive dose-response saturates with a dox power clearly different from $3.4$, the rate-factorization mechanism $k_{\\rm ON}=k_1 k_a$ is wrong.","supporting_citations":[{"cited_title":"Transcriptional control: rheostat converted to on/off switch","cited_arxiv_id":null,"evidence_quote":"Supplies the experimental dox dose-response data and the observed Hill coefficients 1.6, 1.8, and 3.2 that the paper's parameter estimation must fit."},{"cited_title":"Noise in eukaryotic gene expression","cited_arxiv_id":null,"evidence_quote":"Supplies the GAL/aTc-dependent rate-constant forms and the interpretation of k4 as RNAP-II binding tightness used in the stochastic noise analysis."},{"cited_title":"Conversion of graded to binary response in an activator-repressor system","cited_arxiv_id":null,"evidence_quote":"Provides the three-state activator-repressor model of graded-to-binary response whose probability distribution approach the paper extends to full kinetics and noise."},{"cited_title":"Stochastic modeling for the expression of a gene regulated by competing transcription factors","cited_arxiv_id":null,"evidence_quote":"The earlier stochastic Markov chain model that left the Hill-coefficient mismatch and the kinetics unexplained, motivating the present theory."},{"cited_title":"Stochastic gene transcription with non-competitive transcription regulatory architecture","cited_arxiv_id":null,"evidence_quote":"The non-competitive activator-repressor architecture that serves as the comparison baseline for mean expression and Fano factors."},{"cited_title":"Exact stochastic simulation of coupled chemical reactions","cited_arxiv_id":null,"evidence_quote":"The stochastic simulation algorithm used to verify the analytical dose-response curves against the experimental data."},{"cited_title":"Control of noise in gene expression by transcriptional reinitiation","cited_arxiv_id":null,"evidence_quote":"The reinitiation noise-control theory whose treatment of transcriptional reinitiation the present paper incorporates and extends."}],"review_version":1}