{"id":"17dec339-4369-4892-a08f-133f8f65e506","arxiv_id":"2411.17158","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A frequency-to-amplitude converter architecture lets light-pulse frequency, not just intensity, control gene expression and expands the reachable combinations of multiple genes.","lead":"This paper builds a synthetic gene circuit in bacteria that turns the frequency of light pulses into gene expression levels, like an FM receiver decoding a signal. It reports that frequency control can reach combinations of multiple protein outputs that constant light inputs cannot.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"State-space expansion counts are computed from fitted Eq. 6 with an arbitrary epsilon=0.1 grid, not measured directly; the multi-gene experiments never compare AM- vs FM-reachable states, so 'unreachable' is not yet experimentally demonstrated.","rationale":"The most load-bearing concern is not the temporal-decoupling assumption flagged as the reader's weakest_assumption, although that is a legitimate modeling condition. The theory itself holds up internally: the analytical solution and CRN agreement are convincing, and a direct check of the integration leading to Eq. 6 confirms the formula is consistent. The decisive vulnerability is that the paper's central quantitative claim, expanded accessible cellular states under FM, is computed from fitted parameters and an arbitrary discretization, and the multi-gene experiments do not include an experimental AM-only state-space comparison. The reader's rationale correctly identified the fitted-parameters-as-predictions problem and the arbitrary resolution parameter, so there is partial agreement. However, the formal weakest_assumption field pointed to the timescale hierarchy, which I would not select as the single load-bearing issue. My recommended verdict remains CONDITIONAL: the architecture, platform, and theoretical framework are substantial and promising, but the headline claim of accessing states unreachable by amplitude modulation needs direct, independent experimental verification that does not reuse fitted curves or an arbitrary grid. This does not change the reader's conditional verdict, hence UNCHANGED.","tokens_in":19779,"tokens_out":11549,"duration_ms":102515,"concrete_test":"Run a matched experimental accessibility screen on a 2-gene TRGC strain: use the automated platform to map the AM-reachable set by sweeping light intensity I and duty cycle D at a fixed reference frequency (or continuous light), and map the FM-reachable set by sweeping frequency f across the same I,D grid. Define experimentally distinguishable states by a noise-based criterion (e.g., replicate 2-SD separation in (YA,YB) space) instead of epsilon=0.1. Count states in each set and explicitly check whether each claimed FM-only state lies outside the 2-SD neighborhood of every AM-reachable state. If the FM-only increment collapses or the FM-only states fall within the AM set's noise neighborhood, the 'unreachable' claim fails; if a substantial separated set remains, the central claim is confirmed without relying on fitted Eq. 6 or an arbitrary grid resolution.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's headline claim rests on the quantitative state counts (19->38 for two genes, 27->95 for three genes) and on the assertion that FM reaches states 'unreachable through conventional amplitude modulation.' These counts are not directly measured. They are produced by fitting Equation 6 to experimental response curves, then binning the fitted normalized outputs on a 10x10 (or 10^3) grid with an arbitrarily chosen discretization parameter epsilon = 0.1. The main text concedes this grid count is 'primarily a qualitative measure.' Figure 4d/e displays fitting curves through the data rather than independent, noise-resolved state measurements, and no AM-only state set is experimentally mapped for comparison. Because Equation 6 contains four fitted parameters and because 'FM-only' states are defined by grid cells crossed by fitted curves, the 2-fold and 3.5-fold expansions, and especially the word 'unreachable', are not yet supported by independent evidence. The internal theory is strong: I checked the integration leading to Eq. 6 and the apparent missing logarithmic term cancels against the duty-cycle term via (1-sH)/(1-sL)=exp(-phi*D), so the model is internally coherent. The vulnerability is specifically that the central biological claim is validated only through reuse of fitted parameters, not through held-out or direct experimental accessibility tests.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a Time-Resolved Gene Circuit (TRGC) architecture, built on a Frequency-Amplitude Converter (FAC), to implement frequency-modulated gene expression in engineered bacteria. The theoretical core is an analytical solution (Eq. 6) for the steady-state output of a three-module system comprising a wave converter, a Hill-type thresholding filter, and an integrator, which is decomposed into amplitude, duty-cycle, and frequency components (Eqs. 8-10). The authors derive a phase boundary D = 3s*^2 separating high-pass and low-pass regimes, validate the analytical model against CRN simulations (R^2 = 0.992), and report an automated platform that supports frequency-response characterization of 29 engineered P. aeruginosa strains. The final section claims that frequency modulation expands the accessible multi-gene expression state space from 19 to 38 states in two-gene systems and from 27 to 95 states in three-gene systems, thereby reaching states 'unreachable through conventional amplitude modulation.'","tokens_in":20038,"tokens_out":3221,"duration_ms":33155,"significance":"If fully substantiated, the work would be a meaningful advance in synthetic biology: it provides a tractable analytical framework for frequency-to-amplitude conversion, demonstrates high-pass and low-pass filtering in engineered bacteria, and introduces an automated experimental platform that is valuable for dynamic circuit characterization. The analytical derivation is internally coherent, as checked against the CRN simulations, and the authors explicitly acknowledge that their grid-based state counting is a qualitative measure limited by cellular noise. However, the headline claim about expanded cellular states rests on model-based counting using parameters fitted to the same experimental data, not on direct measurements of reachable states, and the paper does not experimentally compare amplitude-modulation-only versus frequency-modulation state sets. This gap prevents the central claim, as currently stated, from being accepted as established.","major_comments":[{"comment":"The central claim that frequency modulation expands accessible states from 19 to 38 (two genes) and from 27 to 95 (three genes) is not supported by direct experimental measurement. These counts are produced by fitting Eq. 6 to experimental response curves (Data Analysis and Supplementary Note 11 state that parameters such as lambda and gamma are determined by fitting), then binning the fitted curves on a grid with an arbitrarily chosen discretization epsilon = 0.1. The main text itself calls this grid-based quantification 'primarily a qualitative measure.' Moreover, Fig. 4d and Fig. 4e display fitting curves through the data rather than independently resolved state measurements, and no experiment maps the state set reachable by amplitude modulation alone for comparison. Therefore the words 'unreachable through conventional amplitude modulation' are not experimentally demonstrated; the paper should either provide a direct experimental comparison of AM-reachable and FM-reachable states, or substantially reframe the claim as a model-based prediction.","section":"Frequency Signal Control Expands Multi-gene Expression State Combinations (Fig. 4b-e)"},{"comment":"The closed-form solution and the derived phase boundary D = 3s*^2 depend on the strict temporal decoupling assumption Tc2 << Tc1 << Tc3, with the thresholding filter responding instantaneously and the integrator averaging with no feedback. The manuscript relies on literature timescales for these estimates but does not report direct measurements of Tc1, Tc2, and Tc3 in the engineered P. aeruginosa strains, nor does it test whether Vfr regulates cpdA or bPAC expression, which would introduce feedback and break the modular decomposition. If these timescales are not sufficiently separated in vivo, Eq. 6 and the phase boundary would not describe the actual circuit. The authors should provide experimental evidence for the timescale hierarchy (for example, direct measurements of cAMP waveform shape and Vfr-promoter response kinetics) or explicitly bound the parameter region in which the analytical solution remains valid.","section":"Theoretical Modeling and Analysis of TRGC (Eqs. 4-6, 9)"},{"comment":"The reported correlation between theory and experiment (R^2 = 0.986 in Fig. 3k) is weakened by the fact that the model parameters are fitted to the same experimental data that are then compared with the model. This is not an independent validation of predictive power; it shows internal consistency of the fitting procedure. The authors should either use held-out data (e.g., fitting on a subset of strains or frequencies and predicting the rest), or state clearly which parameters were free and which were fixed a priori for each comparison.","section":"Fig. 3k and Supplementary Note 11"}],"minor_comments":[{"comment":"There is a typo: 'frequency-dependent r esponses' should be 'frequency-dependent responses.'","section":"Abstract"},{"comment":"The sentence 'the F AC creates bridges the gap' contains a verb error; it should be 'the FAC bridges the gap' or 'creates a bridge across the gap.'","section":"Introduction (FAC architecture paragraph)"},{"comment":"The word 'surcose' should be 'sucrose' in the plasmid loss step.","section":"Methods (Construction of Bacterial Strains)"},{"comment":"The caption has an unclosed parenthesis: 'Blue regions indicate high-pass behavior (YHF > YLF, while red regions...' should be 'YHF > YLF), while red regions...'.","section":"Fig. 2b caption"},{"comment":"The caption of Fig. 4c states that the color gradient represents a transition from amplitude-only (blue) to frequency-modulated (red) states, but the text in the main body and Fig. 4b use blue for amplitude-modulation states and red for additional frequency-modulation states; the color semantics should be described consistently.","section":"Fig. 4b-c captions"}],"recommendation":"major_revision","confidential_remarks":"The paper has a solid theoretical core and a useful experimental platform, but the headline quantitative claim about expanded state space is currently based on model fitting rather than direct measurement. This is fixable within the scope of the manuscript by adding a direct experimental comparison of AM-only and FM-accessible states, or by clearly demoting the claim to a model prediction. The lack of such evidence is the main reason I am not recommending acceptance at this stage."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"What's actually new: the FAC/TRGC architecture, the closed-form solution for frequency-to-amplitude conversion, and the D = 3s*^2 phase boundary. I checked the math, and the stress-test note is right: the apparent missing logarithmic term cancels against the duty-cycle term, so Equation 6 is internally coherent. The CRN agreement (R2 = 0.992) is convincing as internal model validation. The automated platform is a real piece of engineering: 65 strains, continuous culture, stable OD, CV below 10%. That will be useful to the field regardless of what happens with the larger claims.\n\nWhere it gets soft: the experimental validation of the headline claim. The paper repeatedly calls fits to measured data 'predictions.' Parameters lambda, gamma, alpha are fitted to the same frequency-response curves that are then shown to correlate with the model—that is circular, not predictive. The state-space expansion counts (19 to 38, 27 to 95) are computed from binned fitted curves using an arbitrary epsilon = 0.1, and the authors themselves admit the grid count is 'primarily a qualitative measure.' More importantly, the multi-gene experiments never actually map an AM-only state set to compare against. So 'unreachable through conventional amplitude modulation' is an interpretation, not a demonstrated experimental fact. The word 'unreachable' overstates what is shown.\n\nThe temporal decoupling assumption (Tc2 << Tc1 << Tc3) is plausible from known timescales, and the authors do not hide the assumption. That part is fine.\n\nWho is this paper for? Synthetic biologists working on dynamic control, and theorists who want a tractable model of frequency processing in gene circuits. The platform paper alone justifies a read. The central theoretical contribution is sound, but the experimental demonstration of expanded cellular states needs harder evidence—either held-out parameter prediction, direct AM-vs-FM state mapping, or a less loaded claim.\n\nRecommendation: yes, send this to peer review. A serious referee will want to push on the fitting/prediction distinction and the state-count methodology, but the theory and platform are worth the time. My own verdict would be conditional: accept with revisions that either strengthen the experimental validation or substantially soften the 'unreachable' language.","headline":"A solid theory paper with a genuinely useful platform; the headline state-expansion claim is real in the model but not yet experimentally proven.","tokens_in":20602,"tokens_out":1291,"would_cite":true,"duration_ms":14120,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Frequency-modulated gene circuits, built from a wave-converter/threshold-filter/integrator architecture, reach roughly twice the two-gene expression states and 3.5 times the three-gene states of amplitude-only control, the paper reports.","keywords":["Frequency-modulated gene circuits","Dynamic signal processing","Time-resolved biological control","Automated biological characterization","Expanded state control","Synthetic biology","Optogenetics","High-pass low-pass filtering"],"falsifier":"Grow a TRGC strain with light pulses of fixed duty cycle and scan frequency while measuring both the instantaneous Vfr-driven promoter activity and steady output; if the promoter response lags the cAMP wave by a time comparable to the period, or if Vfr is found to regulate cpdA or bPAC expression, the predicted high-pass and low-pass curves and the $D=3s^{*2}$ boundary should fail in vivo.","tokens_in":19566,"feed_emoji":"🧬","tokens_out":9639,"duration_ms":80765,"temperature":0.7,"pith_summary":"This paper seeks to establish that frequency, not just intensity, can serve as an independent control variable in engineered gene circuits. It introduces a Time-Resolved Gene Circuit (TRGC) built around a Frequency-Amplitude Converter (FAC): light pulses are first shaped into sawtooth cAMP waves, then passed through a Vfr-based threshold filter, then integrated into steady protein output. The authors derive a closed-form expression for the output and predict that the same circuit can act as a high-pass or low-pass filter, with the regime boundary set by $D = 3s^{*2}$. In two-gene and three-gene systems driven by a single input, they report that frequency variation raises the number of accessible expression states from 19 to 38 and from 27 to 95, respectively. If correct, this gives synthetic biologists a simple, scalable way to coordinate many genes through one dynamic signal without adding many genetic parts.","feed_headline":"Frequency control doubles the states a gene circuit can reach","feed_subtitle":"A three-module circuit turns light-pulse frequency into gene output, lifting two-gene accessible states from 19 to 38.","key_machinery":"The load-bearing object is the three-module Frequency-Amplitude Converter (FAC), implemented as the Time-Resolved Gene Circuit (TRGC) in Pseudomonas aeruginosa: a Wave Converter (M1, bPAC and CpdA) turns light pulses into a sawtooth cAMP waveform; a Thresholding Filter (M2, Vfr binding) acts as the frequency-selective step; and an Integrator (M3, promoter-driven protein production) averages the filtered signal into steady output. The whole derivation works because the three modules operate on separated timescales, $T_{c2}\\ll T_{c1}\\ll T_{c3}$, so each stage can be analyzed independently and then combined into the closed-form output expression. The threshold value $s^*$ and the resulting phase boundary $D=3s^{*2}$ are what let the same architecture switch between high-pass and low-pass behavior.","core_discovery":"The central claim, stated on the paper's own terms, is that true frequency modulation is a distinct information-encoding channel for synthetic circuits and that the Time-Resolved Gene Circuit realizes it. The Frequency-Amplitude Converter splits input processing into a Wave Converter (M1) that converts square light pulses into sawtooth cAMP oscillations with peak and trough levels $s_H$ and $s_L$; a Thresholding Filter (M2) whose Hill-type activation function has threshold $s^* = 1/\\sqrt{3(\\lambda+1)\\alpha}$; and an Integrator (M3) that time-averages promoter activity. The steady-state output factorizes as $\\bar{y} = y^*(D + G)$, separating a pure amplitude response $y^*$, a duty-cycle term $D$, and a frequency-dependent term $G$, which is what lets one architecture implement both high-pass and low-pass filtering. Experimentally, the paper reports that frequency modulation expands the reachable expression-state grid from 19 to 38 states for two genes and from 27 to 95 for three genes, using an automated continuous-culture platform that keeps cell growth state stable.","pith_inferences":["A natural extension is to transplant the same frequency-to-amplitude conversion into other organisms using any fast reversible activator paired with a slow output protein, since only the timescale ordering matters, not the specific cAMP/Vfr chemistry.","The reported state counts depend on the chosen 0.1 discretization grid, so the durable statement is the ratio of expansion (about 2-fold for two genes, 3.5-fold for three), not the absolute numbers.","Because frequency and intensity are independent input dimensions, frequency control could in principle be layered onto existing amplitude-based circuits to send two signals through one channel, a multiplexing use the paper does not demonstrate.","The $D=3s^{*2}$ boundary doubles as a practical design rule: flipping the duty cycle should flip a low-pass circuit into a high-pass one without rebuilding molecular parts."],"forward_implications":["A single light input controlling several genes can encode extra information in pulse frequency, expanding two-gene expression states from 19 to 38 and three-gene states from 27 to 95 under the paper's 0.1 resolution grid.","The same circuit can act as a tunable high-pass or low-pass filter, with the boundary between regimes set by the simple relation $D = 3s^{*2}$ between duty cycle and threshold.","Frequency response can be tuned at two levels at once: molecular parameters ($\\alpha,\\lambda$) set the operating point, while light intensity, duty cycle, and frequency adjust the response online, with high-pass configurations giving larger frequency discrimination than low-pass ones.","Because the expansion grows with the number of regulated genes, frequency control becomes more valuable as the network gets larger, offering a path to coordinate many genes from a single dynamic input."],"supporting_citations":[{"why":"Supplies the natural example of signal-dependent transcription-factor translocation that motivates frequency-based regulation.","marker":"[2]"},{"why":"Provides the general encoding and decoding framework for signaling dynamics that the paper extends to synthetic circuits.","marker":"[3]"},{"why":"Documents frequency-modulated nuclear-localization bursts as a natural precedent for FM gene regulation.","marker":"[5]"},{"why":"Distinguishes functional roles of pulsing and defines why pure frequency modulation differs from pulse-width modulation.","marker":"[7]"},{"why":"Establishes pulse-width modulation of optical signals as the prior synthetic approach that the TRGC goes beyond.","marker":"[27]"},{"why":"Shows that pulsatile inputs can achieve graded multi-gene regulation, the key baseline for the claimed state-space expansion.","marker":"[28]"},{"why":"Supplies the control-theory perspective on modular circuit design and timescale separation used to build the TRGC.","marker":"[38]"},{"why":"Characterizes bPAC's light-controlled cAMP production, the kinetic basis for the Wave Converter timescale.","marker":"[41]"},{"why":"Provides the synthetic-gene-network context for arguing that the FAC needs only a modest genetic architecture.","marker":"[47]"},{"why":"Quantifies optimal frequencies for cAMP information transmission in bacteria, informing the experimental frequency window.","marker":"[55]"}],"fun_headline_variants":["Frequency-modulated gene circuits double reachable cell states","Light-pulse frequency expands gene circuit states from 19 to 38","Time-Resolved Gene Circuit unlocks new cellular expression states","Synthetic circuits read frequency, not amplitude, to expand states","Frequency control in gene circuits expands expression space two-fold"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the three modules act on well-separated timescales, with the Vfr threshold filter responding instantly to cAMP, the cAMP wave converter operating on an intermediate timescale, and the output integrator averaging slowly, and with no feedback from the output back into the earlier modules, so if these timescales overlap in the real bacterium or Vfr feeds back onto cAMP production or degradation, the closed-form solutions and the $D=3s^{*2}$ boundary need not hold in vivo.","fun_headline_variants_meta":{"raw":{"variants":["Frequency-modulated gene circuits double reachable cell states","Light-pulse frequency expands gene circuit states from 19 to 38","Time-Resolved Gene Circuit unlocks new cellular expression states","Synthetic circuits read frequency, not amplitude, to expand states","Frequency control in gene circuits expands expression space two-fold"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00019,"raw_usage":{"total_tokens":1363,"prompt_tokens":992,"completion_tokens":371,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":608,"completion_tokens_details":{"reasoning_tokens":289}},"tokens_in":608,"tokens_out":371,"duration_ms":4090,"temperature":1.0,"reasoning_tokens":289,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T12:28:43.352202+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Grow a TRGC strain with light pulses of fixed duty cycle and scan frequency while measuring both the instantaneous Vfr-driven promoter activity and steady output; if the promoter response lags the cAMP wave by a time comparable to the period, or if Vfr is found to regulate cpdA or bPAC expression, the predicted high-pass and low-pass curves and the $D=3s^{*2}$ boundary should fail in vivo.","supporting_citations":[{"cited_title":"Nature structural & molecular biology 19(1), 31–39 (2012)","cited_arxiv_id":null,"evidence_quote":"Supplies the natural example of signal-dependent transcription-factor translocation that motivates frequency-based regulation."},{"cited_title":"Cell 152(5), 945–956 (2013)","cited_arxiv_id":null,"evidence_quote":"Provides the general encoding and decoding framework for signaling dynamics that the paper extends to synthetic circuits."},{"cited_title":"Nature 455(7212), 485–490 (2008)","cited_arxiv_id":null,"evidence_quote":"Documents frequency-modulated nuclear-localization bursts as a natural precedent for FM gene regulation."},{"cited_title":"Science 342(6163), 1193–1200 (2013)","cited_arxiv_id":null,"evidence_quote":"Distinguishes functional roles of pulsing and defines why pure frequency modulation differs from pulse-width modulation."},{"cited_title":"Journal of molecular biology 425(22), 4161–4166 (2013)","cited_arxiv_id":null,"evidence_quote":"Establishes pulse-width modulation of optical signals as the prior synthetic approach that the TRGC goes beyond."},{"cited_title":"Nature communications 9(1), 3521 (2018)","cited_arxiv_id":null,"evidence_quote":"Shows that pulsatile inputs can achieve graded multi-gene regulation, the key baseline for the claimed state-space expansion."},{"cited_title":"Journal of The Royal Society Interface 13(120), 20160380 (2016)","cited_arxiv_id":null,"evidence_quote":"Supplies the control-theory perspective on modular circuit design and timescale separation used to build the TRGC."},{"cited_title":"Journal of Biological Chemistry 286(2), 1181–1188 (2011)","cited_arxiv_id":null,"evidence_quote":"Characterizes bPAC's light-controlled cAMP production, the kinetic basis for the Wave Converter timescale."},{"cited_title":"Nature biotechnol- ogy 27(12), 1139–1150 (2009)","cited_arxiv_id":null,"evidence_quote":"Provides the synthetic-gene-network context for arguing that the FAC needs only a modest genetic architecture."},{"cited_title":"Optimal Frequency in Second Messenger Signaling Quantifying cAMP Information Transmission in Bacteria","cited_arxiv_id":"2408.04988","evidence_quote":"Quantifies optimal frequencies for cAMP information transmission in bacteria, informing the experimental frequency window."}],"review_version":1}