{"id":"ff917793-e5f2-469f-a365-0800ca3093aa","arxiv_id":"2507.19395","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"In a stochastic ribosome-traffic model with antibiotic-induced pauses and finite mRNA lifetimes, longer transcripts suffer disproportionately stronger translation inhibition, driven by collective ribosome dynamics rather than single-ribosome kinetics.","lead":"Ribosome-stalling antibiotics like chloramphenicol cut protein output far more on long genes than on short ones, because a longer transcript holds more ribosomes and is therefore much more likely to suffer a traffic-blocking stall before the mRNA falls apart. The model suggests cells can blunt this vulnerability by lowering the rate at which ribosomes start translating a given message.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (38) treats the random first-pause time as its mean T0; this unvalidated mean-field approximation, not the Single-Cluster Approximation, is the load-bearing step for the quantitative match in Fig. 8.","rationale":"The reader's weakest assumption is the Single-Cluster Approximation (Sec. II B, III B). I agree that approximation is underquantified, but it is not the most load-bearing step for the paper's central claims. The length dependence (i), initiation-rate mitigation (ii), and collective-vs-single (iii) are derived from the first-pause time T0 and the hitting probability (Eqs. (15), (32)), not from Tp or the jam-dissolution formulas. The experimental validation in Fig. 8 uses Eq. (38), which depends only on T0, t_L, J0, and fitted alpha and tau. The SCA enters only through Tp and the density/current expressions (Figs. 4-6), which are not needed for the experimental ratio. The central quantitative claim therefore hinges on the treatment of T0 as deterministic. The paper validates T0 itself against simulations (Fig. 2), but never validates the protein-output average that converts T0 into P. The exact expectation is easily computed from the same Poisson-process ingredients, and the difference is likely to matter because the coefficient of variation of the first-pause time is of order one for these parameters, and because the hard cutoff T0 <= t_L creates a discontinuity in predicted expression. This is a concrete, testable mathematical issue rather than a vague regime concern. The reader's CONDITIONAL verdict remains appropriate because the model's qualitative structure may still be right, but the specific quantitative match should not be taken as established until the averaging check is performed.","tokens_in":17856,"tokens_out":12489,"duration_ms":143640,"concrete_test":"Simulate the full Gillespie model with the same parameters as Fig. 8 (alpha=0.12 s^-1, tau=170 s, kon, ku, epsilon=20 s^-1, l=10, L=238 and 1024) but including stochastic mRNA degradation with rate 1/tau; compute the mean protein output and the ratio R as a function of chloramphenicol concentration, and compare with Eq. (38). If the mean-replacement error in R is small (e.g., <10%), the concern is resolved; if it is large, the quantitative match in Fig. 8 is an artifact and the central claim requires re-analysis.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The quantitative agreement claimed in the abstract and shown in Fig. 8 rests on the protein-output formula Eq. (38) in Sec. IV C. There, T0 is defined as the expected lifetime of the unpaused state (Eq. (15)), but Eq. (37) uses it as if every mRNA either degrades by T0 or pauses exactly at T0. The actual first-pause time has the broad distribution F(t) in Eq. (11); the correct expectation over F(t) and the mRNA lifetime distribution (Eq. (33)) is an integral over the stopping rate lambda(t)+1/tau weighted by exp(-Lambda(t)-t/tau). The mean replacement is nowhere validated against Gillespie simulations, and it introduces a hard cutoff: Eq. (38) sets P=0 whenever T0 <= t_L. For lacZ at high chloramphenicol concentrations (where kp approaches or exceeds ku), T0 approaches t_L, so the formula can predict zero output for the long gene while the stochastic process would still produce finite protein from mRNAs whose actual pause time exceeds t_L, and from ribosomes that have already passed the pause site. This could steepen the short-to-long ratio R and create the appearance of quantitative agreement.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a TASEP model of ribosome translation with stochastic antibiotic-induced pausing and unpausing, and introduces a Single-Cluster Approximation in which the first pause creates a single contiguous jam that fills back to the initiation site and dissolves in discrete batches. The authors derive analytical expressions for the unpaused-state lifetime, first-pause position, paused-state lifetime, density, current, particle state fractions, and hitting probability, and validate these against Gillespie simulations. They then extend the model to include finite mRNA lifetimes and use it to predict the short-to-long expression ratio for gfp and lacZ under chloramphenicol, reporting quantitative agreement with the experimental data of Zhang et al. (Fig. 8). The central claims are that antibiotic-induced inhibition is strongly transcript-length dependent, that lowering initiation rates mitigates vulnerability, and that collective ribosome dynamics, rather than single-ribosome properties, govern the response.","tokens_in":18109,"tokens_out":6750,"duration_ms":81649,"significance":"If the results hold, this is a valuable contribution to the modeling of translation under antibiotic stress. The paper's strengths are its explicit analytical treatment, the use of experimentally measured pausing/unpausing rates, and the systematic Gillespie validation of the core observables (lifetimes, densities, currents, and fractions) in Figs. 2-7. The length-dependence prediction and the initiation-rate mitigation effect are concrete and falsifiable, and the model gives a mechanistic explanation for why longer transcripts are more vulnerable. The comparison with Zhang et al. is an ambitious step toward quantitative validation, but it is the least secure part of the paper because it relies on an unvalidated mean-field replacement and on fitted parameters whose number is reported inconsistently. The core modeling framework is nevertheless defensible and the analytical results for the pre-validation part are credible.","major_comments":[{"comment":"The protein-output formula replaces the random first-pause time by its mean T0. The actual first-pause time has the broad distribution F(t) in Eq. (11), and mRNA degradation is exponential with density in Eq. (33); the correct expected output is an integral over the stopping time min(T_pause, T_degradation), not an evaluation at the mean T0. This mean replacement is never validated against Gillespie simulations, and it creates a hard cutoff: Eq. (38) sets P=0 whenever T0 <= tL. For lacZ at high chloramphenicol concentrations, e.g. kp ~ 0.027 s^-1 with the parameters in Fig. 8, one obtains T0 ~ 23 s while tL ~ 51 s, so Eq. (38) predicts zero lacZ output even though stochastic realizations with first-pause times beyond tL, and ribosomes that have already passed the pause site, would still produce finite protein. This could artificially steepen the short-to-long ratio R and create the appearance of quantitative agreement. This is the load-bearing step for the central validation claim, so it must be corrected by integrating over F(t) and the mRNA lifetime distribution, or otherwise rigorously justified against simulations, and the impact on Fig. 8 must be reassessed.","section":null},{"comment":"The validity conditions of the Single-Cluster Approximation are stated only qualitatively: kp << epsilon, beta ~ epsilon, and 'the segment cannot be too long'. The approximation is used in Fig. 8 at antibiotic concentrations where kp is not small compared to ku; for example, at 10 uM chloramphenicol kp ~ 5.4e-3 s^-1 and ku = 1.4e-3 s^-1, so the batch size B = ku/kp + 1 is about 1.26, meaning the 'clusters' are small and multiple pauses can occur before a cluster dissolves. The authors should quantify the regime of validity, ideally with a phase diagram or simulation-based boundary, and demonstrate that the experimental comparison in Fig. 8 lies inside that regime. Without this, the analytical formulas for Tp, density, and current, and therefore the protein-output prediction, are not guaranteed to apply in the very regime used for the quantitative claim.","section":null},{"comment":"The number of fitted parameters in the experimental comparison is reported inconsistently. The Fig. 8 caption states that a fitted initiation rate alpha = 0.12 s^-1 and a mean mRNA lifetime tau = 170 s are used, while the Discussion states that the model 'quantitatively reproduces experimental differences ... by fitting only the degradation time tau'. This discrepancy matters because the 'quantitative agreement' is not parameter-free: it depends on two biologically plausible but adjustable parameters, and no confidence intervals, sensitivity analysis, or goodness-of-fit measure is reported. The authors should clarify which parameters were fitted, show how the fit was performed, and report the sensitivity of the short-to-long ratio R to alpha and tau, especially given the hard cutoff issue in Eq. (38).","section":null}],"minor_comments":[{"comment":"The quantity tL is defined in Eq. (5) as the domain-wall traversal time L/(epsilon-alpha), but in Sec. IV B it is called 'the time required for a ribosome to translate the full mRNA', which for a single ribosome should be L/epsilon. This notational reuse should be clarified, since the comparison with the single-ribosome hitting probability relies on the same symbol.","section":null},{"comment":"The binding rate constant kon is given as 5.6 x 10^-4 uM^-1 s^-1 in the text and 5.4 x 10^-4 uM^-1 s^-1 in the Fig. 8 caption. Please reconcile the values and cite the corresponding uncertainty.","section":null},{"comment":"Some figure labels appear truncated or orphaned: Fig. 4 contains a floating heading 'Paused-state fraction,' and the legend entries in Fig. 6 are cut off ('(mobile+jammed)' and 'this work (mobile)'). The figures should be readable without referring to the main text.","section":null},{"comment":"The text has a minor typo ('mRNAS' for 'mRNAs'), and the notation P(t) for protein number is easily confused with the probability P0 and Pp from Eq. (21). A different symbol or a brief reminder would improve readability.","section":null}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope and the analytical modeling is mostly careful, but the quantitative validation against the Zhang et al. data is the least reliable part. The mean-field replacement in Eq. (38) is not merely a cosmetic approximation: it can produce exactly zero output for the long gene in the high-concentration regime, which would strongly bias the reported ratio R. I would ask the authors to redo the protein-output calculation by integrating over the true first-pause time distribution, or to demonstrate numerically that the mean replacement is accurate in the parameter range of Fig. 8. If the corrected comparison weakens the quantitative agreement, the abstract and conclusion should be softened accordingly. The inconsistency between the Discussion ('fitting only tau') and the Fig. 8 caption ('fitted initiation rate') should also be fixed, since it affects how readers assess the number of free parameters."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Johannes,\n\nIf you only have time for one thing: the paper's central mechanism is real and its analytics mostly check out, but the 'quantitative agreement' in Fig. 8 rests on a mean-field replacement of a random first-pause time that is not validated anywhere in the paper.\n\nWhat is genuinely new: the combination of finite-length TASEP, stochastic pausing, and explicit mRNA degradation, analyzed via a Single-Cluster Approximation, is a fresh and biologically motivated package. The prediction that longer transcripts are disproportionately vulnerable, and that lowering initiation rates mitigates that vulnerability, is a clean, testable consequence of collective ribosome traffic rather than an input. The analytical expressions match the paper's own Gillespie simulations across a broad parameter range—that is real evidence of internal consistency. The paper also makes a fair point against the single-ribosome hitting probability in Eq. (31); the collective version in Eq. (32) is simple and correct in spirit.\n\nNow the soft spots, in proportion. The load-bearing issue is Eq. (38) in Sec. IV C. The derivation treats T0—the expected unpaused lifetime—as if every mRNA either degrades by T0 or pauses exactly at T0. The actual first-pause time F(t) has a broad distribution, and the correct expectation is an integral over that distribution weighted by the mRNA lifetime distribution. The paper never tests this mean replacement against simulations, and it creates a hard cutoff: for lacZ at high chloramphenicol concentrations, T0 approaches t_L, so the formula can predict zero output while the stochastic process still produces protein from mRNAs whose real pause time is later. That could easily steepen the short-to-long ratio R and manufacture quantitative agreement. This is a real flaw, not a quibble.\n\nSecond, the experimental validation in Fig. 8 uses two fitted parameters (alpha = 0.12 s^-1, tau = 170 s), and no competing single-ribosome model is fit to the same data, so the 'collective dynamics govern' claim is not yet tested against a simpler alternative. Third, the Single-Cluster Approximation's validity conditions—kp much less than epsilon, beta approximately epsilon, lattice not too long—are stated qualitatively and never quantified; the regime used for the experimental comparison is exactly where the approximation is most strained.\n\nAll that said, the central length-dependence direction is robust, the paper is honest about its limitations, and the framework is a useful stepping stone. The typos are fixable. This paper deserves a serious referee; the referee should ask for a simulation-based check of Eq. (38) and a comparison against a single-ribosome baseline.\n\nBest,\n[Your name]","headline":"The paper's core length-dependence mechanism is real and the analytics largely hold up, but the Fig. 8 quantitative match rests on an unvalidated mean-field replacement of the random pause time.","tokens_in":18611,"tokens_out":1165,"would_cite":true,"duration_ms":14891,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Ribosome-targeting antibiotics inhibit longer mRNAs more than short ones, and reducing ribosome initiation offsets the damage.","keywords":["Totally Asymmetric Simple Exclusion Process (TASEP)","ribosome pausing","ribosome-targeting antibiotics","chloramphenicol","transcript length dependence","mRNA degradation","translation inhibition","collective ribosome dynamics"],"falsifier":"Watch a single mRNA under fluorophore-tagged ribosomes with chloramphenicol at concentrations where $k_p$ approaches $k_u$: if a second spatially separate ribosome cluster appears before the first one dissolves, or if the measured first-pause time and jam-clearing time deviate from $T_0$ and $T_p$, the central mechanism fails. A population-level test is equally direct: measure fold-change expression of a long and a short reporter driven by identical promoters over a range of antibiotic concentrations and initiation rates; the model's short-to-long ratio must rise with concentration and flatten when initiation is lowered.","tokens_in":17615,"feed_emoji":"🧬","tokens_out":6219,"duration_ms":67166,"temperature":0.7,"pith_summary":"The paper sets out to show that the cellular impact of ribosome-targeting antibiotics cannot be read off from what happens to a single ribosome: because ribosomes queue on an mRNA, one antibiotic-induced pause nucleates a traffic jam that grows backward toward the start codon, so the damage scales with transcript length and with how often new ribosomes start. The authors build a TASEP with stochastic pausing and experimentally measured binding and unbinding rates, and they introduce a Single-Cluster approximation that yields closed-form expressions for the time until the first pause, the time to dissolve the resulting jam, and the resulting density and protein-synthesis current. Their analytical results match Gillespie simulations, and the model quantitatively reproduces the measured short-to-long expression ratio for gfp and lacZ under chloramphenicol. A sympathetic reader would care because the model turns a generic \"antibiotics stall ribosomes\" statement into concrete, testable predictions: long genes are more vulnerable, and slowing initiation protects them.","feed_headline":"Longer mRNAs are hit hardest by ribosome-stalling antibiotics","feed_subtitle":"A ribosome-traffic model reproduces chloramphenicol's short-vs-long expression gap and points to initiation rate as a shield.","key_machinery":"The carrying object is the Single-Cluster Approximation: after the first antibiotic-induced pause at position $X_f$, a single contiguous cluster of ribosomes forms and extends back to the initiation site, then dissolves in discrete batches of average size $B = \\frac{k_u}{k_p}[1-(\\frac{k_p}{k_p+k_u})^N] + 1$ before any additional distinct cluster appears. This approximation splits the dynamics into two states: an unpaused state obeying standard TASEP with extended particles, and a paused state whose lifetime is $T_p = n_B/k_u$ with $n_B = \\lceil N_i/B \\rceil$ batches to clear. All analytical expressions for density, current, and the fractions of paused, jammed, and mobile ribosomes follow from the probabilities $P_0$ and $P_p$ of being in each state, making the approximation the bridge from a many-body exclusion process to closed-form formulas.","core_discovery":"The central discovery is that translation inhibition by elongation-stalling antibiotics is a collective, length-dependent phenomenon. In the model, an antibiotic molecule binds a ribosome at rate $k_p$, pausing it for an average time $1/k_u$ that is long compared with elongation, mRNA lifetime, and initiation intervals. That single paused ribosome blocks all ribosomes behind it, forming one contiguous cluster that reaches back to the initiation site; the jam then dissolves in discrete batches as paused ribosomes unpause. Because longer transcripts hold more ribosomes at steady state, they are hit sooner, their jam takes longer to clear, and their protein output per mRNA drops more steeply with antibiotic concentration. The model's key derived quantities, the unpaused-state lifetime $T_0$, the paused-state lifetime $T_p$, and the density $\\rho$ and current $J$ as weighted averages of the two states, reproduce Gillespie simulations and, when mRNA degradation is added, the experimental gfp/lacZ expression ratio under chloramphenicol.","pith_inferences":["The paper's mechanism implies a design principle: cells can buffer long genes against stochastic pausing by lowering their initiation rates, so the observed anticorrelation between gene length and initiation rate in bacteria and yeast may be a consequence of such buffering rather than an unrelated trend.","Because the first-pause time is set by the collective number of ribosomes on the transcript, reporter-gene comparisons under antibiotics should control for initiation rate as well as length; otherwise differences in ribosome loading would masquerade as length effects.","A single-molecule translation assay should reveal a bimodal protein-output distribution per mRNA under antibiotic stress: either a near-normal yield if the message degrades before the first pause, or near zero if it is hit, with the fraction of zero-yield transcripts growing with transcript length.","At high antibiotic binding rates the Single-Cluster Approximation is expected to fail, but the length-dependence should persist; an extension treating coagulation and decoagulation of multiple clusters could sharpen the quantitative predictions precisely in the regime where the analytical formulas start to break down.",""],"forward_implications":["If correct, chloramphenicol-like antibiotics should reduce protein output per mRNA more for long genes than for short genes even when promoter strength, mRNA abundance, and initiation and elongation rates are identical.","Lowering ribosome initiation rate should measurably reduce a transcript's antibiotic vulnerability, by lengthening the time to the first pause and shrinking the jam that forms once a pause occurs.","Estimates of the active-ribosome fraction that count only paused versus non-paused ribosomes overstate protein synthesis under antibiotic stress, because jammed but unpaused ribosomes contribute to density but not to current.","The short-to-long expression ratio $R$ should rise monotonically with antibiotic concentration, and the model gives a quantitative curve for that rise once initiation rate, mRNA lifetime $\\tau$, $k_{\\text{on}}$, and $k_u$ are fixed.","The same two-state pausing-TASEP picture extends to other driven biological traffic with rare, long pauses, notably transcription by RNA polymerases.",""],"supporting_citations":[{"why":"Introduces the TASEP as a model of ribosome traffic; supplies the base exclusion process the paper extends with stochastic pausing.","marker":"[3]"},{"why":"Provides the extended-particle TASEP mean-field expressions for current and density used for the unpaused state.","marker":"[9]"},{"why":"Supplies the experimental gfp and lacZ expression data under antibiotic treatment that the model reproduces in Fig. 8.","marker":"[20]"},{"why":"Provides the measured chloramphenicol binding and unpausing rates $k_{\\text{on}}$ and $k_u$ used to set model parameters.","marker":"[22]"},{"why":"Gives physiological E. coli elongation and initiation rates and the single-ribosome hitting probability formula the paper generalizes to collective dynamics.","marker":"[26]"},{"why":"Supplies the mRNA length distribution, average mRNA lifetime, and the gene-length/initiation-rate observations used in calibration and in the RNAP extension.","marker":"[29]"},{"why":"The predecessor finite-size driven lattice gas with pausing and dynamical defects from which the batch-size geometric result and dynamic-obstacle mapping are taken.","marker":"[30]"},{"why":"The Gillespie algorithm used for the stochastic simulations that validate all analytical predictions.","marker":"[49]"}],"fun_headline_variants":["Longer mRNAs take the biggest hit from stalling antibiotics","Ribosome jams make stalling antibiotics worse for long genes","Lower ribosome initiation rate shields genes from antibiotic stalls","Antibiotic stall impact on translation scales with mRNA length","Collective ribosome jams set antibiotic sensitivity by gene length"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole calculation rests on the Single-Cluster Approximation: after the first antibiotic-induced pause, exactly one contiguous jam forms and dissolves completely before any second distinct cluster can appear, an assumption the paper states is valid for slow-binding antibiotics but never quantifies.","fun_headline_variants_meta":{"raw":{"variants":["Longer mRNAs take the biggest hit from stalling antibiotics","Ribosome jams make stalling antibiotics worse for long genes","Lower ribosome initiation rate shields genes from antibiotic stalls","Antibiotic stall impact on translation scales with mRNA length","Collective ribosome jams set antibiotic sensitivity by gene length"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000587,"raw_usage":{"total_tokens":2788,"prompt_tokens":1009,"completion_tokens":1779,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":625,"completion_tokens_details":{"reasoning_tokens":1696}},"tokens_in":625,"tokens_out":1779,"duration_ms":14971,"temperature":1.0,"reasoning_tokens":1696,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T14:21:49.435260+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Watch a single mRNA under fluorophore-tagged ribosomes with chloramphenicol at concentrations where $k_p$ approaches $k_u$: if a second spatially separate ribosome cluster appears before the first one dissolves, or if the measured first-pause time and jam-clearing time deviate from $T_0$ and $T_p$, the central mechanism fails. A population-level test is equally direct: measure fold-change expression of a long and a short reporter driven by identical promoters over a range of antibiotic concentrations and initiation rates; the model's short-to-long ratio must rise with concentration and flatten when initiation is lowered.","supporting_citations":[{"cited_title":"To this end we first consider the instantaneous position of the domain wall, xs(t), as de- fined in Eq","cited_arxiv_id":null,"evidence_quote":"Introduces the TASEP as a model of ribosome traffic; supplies the base exclusion process the paper extends with stochastic pausing."},{"cited_title":"Kavˇ ciˇ c, G","cited_arxiv_id":null,"evidence_quote":"Provides the extended-particle TASEP mean-field expressions for current and density used for the unpaused state."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the experimental gfp and lacZ expression data under antibiotic treatment that the model reproduces in Fig. 8."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the measured chloramphenicol binding and unpausing rates $k_{\\text{on}}$ and $k_u$ used to set model parameters."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives physiological E. coli elongation and initiation rates and the single-ribosome hitting probability formula the paper generalizes to collective dynamics."},{"cited_title":"Chopra and T","cited_arxiv_id":null,"evidence_quote":"The predecessor finite-size driven lattice gas with pausing and dynamical defects from which the batch-size geometric result and dynamic-obstacle mapping are taken."},{"cited_title":"Scott, E","cited_arxiv_id":null,"evidence_quote":"The Gillespie algorithm used for the stochastic simulations that validate all analytical predictions."}],"review_version":1}