{"id":"de713c36-0c3f-4f22-934f-c8a898e99bd7","arxiv_id":"2608.00995","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A two-layer random-walk model of mRNA transport in dendritic trees predicts that a retrograde persistence bias equalizes travel times to synapses, while an anterograde bias makes them grow linearly with distance.","lead":"This paper builds two levels of math for how mRNA molecules travel inside nerve cells, from the cell body to the synapses. It predicts that a slight preference for backward motion makes arrival times at different synapses nearly identical, while a forward preference makes arrival time grow with distance.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central claim rests on a 0.02 persistence-bias difference with no uncertainty; both sign and magnitude of beta-alpha are load-bearing, and the stated asymptotic criterion l >> v/lambda is insufficient.","rationale":"The reader's weakest_assumption correctly identifies the fragility of the persistence-bias difference as the main load-bearing concern. My stress-test refines it by adding the requirement l >> v/(lambda*|beta-alpha|), which makes the concern stronger: even the qualitative regime may fail if the true bias is small, not just the quantitative assignment. The theoretical framework itself appears internally consistent: the single-dendrite PTPP results are checked by Monte Carlo simulation, the semi-Markov construction uses standard absorbing-Markov-chain theory, and the two regimes are plausible asymptotic limits. The absolute-time issue (CMFPT ~1e7 s vs. mRNA half-life) is real but explicitly acknowledged by the authors as a limitation; it does not invalidate the mathematical predictions, only their immediate biological interpretation. Therefore the appropriate disposition remains CONDITIONAL, which is what the reader recommended; my analysis does not change that verdict.","tokens_in":22045,"tokens_out":20883,"duration_ms":180003,"concrete_test":"Re-analyze the data from Song et al. [17] (or extract the published per-molecule traces) to bootstrap a 95% confidence interval for beta-alpha. If the interval straddles zero or includes both signs, the regime assignment for the human and mouse neurons is unsupported. Additionally, using the lower confidence bound |beta-alpha|_low, compute the length scale v/(lambda*|beta-alpha|_low) and compare with the distribution of dendrite lengths in the NeuroMorpho reconstructions; if a substantial fraction of dendrites is shorter than this scale, the asymptotic predictions of Eq. (9) do not apply to those neurons.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's application to real neurons (Fig. 4, abstract's two regimes) depends on the sign of beta-alpha: the quoted values alpha=0.51, beta=0.53 differ by 0.02, with no error bars or significance test from [17]. If the true bias is zero or opposite in sign, the predicted retrograde-bias regime for the human and mouse neurons flips or becomes undefined. The sensitivity is worse than the reader notes: the asymptotic formulas in Eq. (9) are not valid merely for l >> v/lambda. The exponent controlling the regime is k = (beta-alpha)*lambda/v, so the asymptotic condition is l >> v/(lambda*|beta-alpha|). For the quoted difference this crossover is ~11.4 um, but if the true |beta-alpha| is smaller (e.g., 0.001), the crossover exceeds typical dendrite lengths and the predicted flat or linear behavior is not attained. In addition, the retrograde-regime CMFPT values (~6.3e7 s) are orders of magnitude longer than typical mRNA half-life; the paper acknowledges degradation as future work but this makes the biological relevance of the 'equal MFPT' prediction questionable. These are addressable empirical and modeling issues, not internal contradictions: the single-dendrite mathematics is simulation-validated and the semi-Markov formulation is standard.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper develops a two-level analytical framework for mRNA transport in neuronal dendritic trees. At the single-dendrite level, the motion is modeled as a persistent telegraph process with pauses (PTPP), and closed-form expressions are derived for the probability of traversing the dendrite and the conditional mean first-passage time (CMFPT), for both asymmetric and symmetric persistence probabilities. At the whole-neuron level, the dendrite-level quantities are used as transition probabilities and sojourn times in a semi-Markov chain on the dendritic tree, yielding the matrix equation Ω = ND for the weighted MFPTs from the soma to each synapse. Using experimentally estimated parameters from Song et al. [17] and morphologies from NeuroMorpho, the model predicts two regimes: when retrograde persistence is stronger (β > α), CMFPTs to different synapses become nearly equal; when anterograde persistence is stronger (α > β), the CMFPT grows approximately linearly with soma–synapse distance, resembling transport on a single dendrite. The single-dendrite analytical results are validated by Monte Carlo simulations, and the code is publicly available.","tokens_in":22267,"tokens_out":8344,"duration_ms":70077,"significance":"If correct, the paper provides a valuable, analytically tractable bridge from single-filament run-and-pause kinetics to network-level first-passage statistics in neurons. The single-dendrite solutions are derived from first principles, are simulation-validated, and the semi-Markov reduction is standard and cleanly presented. The prediction of two qualitatively different regimes—nearly equal synaptic arrival times versus distance-linear arrival times—is striking and in principle falsifiable. The availability of code and the use of experimentally derived parameters are clear strengths. However, the significance is tempered by two concerns: the regime classification for real neurons depends on a very small and unquantified persistence difference (β−α = 0.02), and the retrograde-regime CMFPT values (≈6×10^7 s) far exceed typical mRNA half-lives, so the biological interpretation of the \"equal MFPT\" prediction requires further justification. These issues are addressable and do not invalidate the mathematical core.","major_comments":[{"comment":"The paper repeatedly states that the asymptotic results are valid for dendrite lengths l >> v/λ, but the exponential rate controlling the asymptotic regime is k = (β−α)λ/v, not λ/v. The linear asymptotics in Eq. (9) therefore require l >> v/(λ|β−α|). With the parameters used in the paper (λ = 4.4 s⁻¹, v = 1 μm/s, β−α = 0.02), this crossover length is about 11.4 μm, not 0.23 μm as the criterion v/λ would suggest. If the true persistence difference were smaller, e.g., |β−α| = 0.001, the crossover would exceed 200 μm, which is larger than many dendrites; the predicted regimes would then not be attained. The stated condition is therefore not merely imprecise but can be wrong by two orders of magnitude, and the theoretical justification of the abstract's two-regime claim needs to be recast using the corrected criterion.","section":"Numerical examples and Eq. (9); Supplement Sec. A5"},{"comment":"The assignment of the human and mouse neurons to the retrograde-bias regime rests entirely on the point estimates α = 0.51 and β = 0.53 taken from [17]. The difference of 0.02 is reported with no uncertainty, no significance test, and no discussion of how sensitive the predicted regime is to this value. The sign of β−α is load-bearing: if the true bias is zero or has the opposite sign, the predicted nearly-equal-MFPT behavior disappears or flips to the linear-regime behavior. Furthermore, as β−α → 0 the asymptotic slopes in Eq. (9) diverge, so the two regimes are not continuously connected and even a modest uncertainty in the persistence probabilities can change the qualitative prediction. I request a sensitivity analysis—for example, recomputing the CMFPT-distance curves for β−α within plausible error bounds, or plotting the regime boundary as a function of β−α—so that the biological claims are robust to the uncertainty in the experimental input.","section":"Numerical examples (α = 0.51, β = 0.53 paragraph)"},{"comment":"The retrograde-regime CMFPT for the human neuron is reported as roughly 6.3×10^7 s (about two years), which is several orders of magnitude longer than typical mRNA half-lives. Because the model conditions on eventual absorption at a synapse, it describes the age of mRNA molecules that actually arrive; but if degradation is taken into account, most mRNA would be degraded long before arrival, and the \"nearly equal MFPT\" regime would concern only a tiny surviving fraction. The paper acknowledges degradation as future work, but the current presentation does not quantify the impact of this model choice on the biological interpretation of the main prediction. I suggest adding a simple degradation cut-off (e.g., conditioning on arrival before a finite time T, or introducing an exponential survival probability) to indicate whether the predicted equal-arrival-time effect would be observable in practice.","section":"Numerical examples (human neuron, bias ~6.3×10^7 s); Conclusions"}],"minor_comments":[{"comment":"The symbol f_ms in Eq. (8) is undefined; from the supplement Eq. (B24) the intended quantity is clearly π_ms, the probability of absorption at synapse s starting from node m.","section":"Main text, Eq. (8)"},{"comment":"The generator equations are displayed as Eq. (3) but the text immediately afterward refers to \"the system (4)\"; the numbering should be consistent.","section":"Main text, Eq. (3) and following paragraph"},{"comment":"Typo: \"CMPFTs\" should be \"CMFPTs\" (conditional mean first-passage times).","section":"Main text, paragraph after Eq. (3)"},{"comment":"The validity condition for the asymptotic results is consistently stated as \"l >> v/λ\"; as noted in major comment 1, the correct criterion is l >> v/(λ|β−α|), and this should be corrected in the abstract and throughout the numerical examples.","section":"Abstract and main text (repeated)"},{"comment":"The y-axis label \"S\" is ambiguous without units; please specify that S denotes the slope of the CMFPT versus length in units of s/μm.","section":"Figure 3(b)"},{"comment":"The constant in Eq. (A56) is written as \"Q2L\"; this appears to be a typo, and the prefactor should be consistent with the symmetric-case constants C5 = C6 in Eq. (A54).","section":"Supplement, Eq. (A56)"},{"comment":"Typo: \"stro0nger\" should be \"stronger\".","section":"Supplement, final paragraph"}],"recommendation":"major_revision","confidential_remarks":"The mathematical core of the paper is sound and the single-dendrite solutions are validated by Monte Carlo. My main reservation is that the central qualitative prediction for real neurons depends on a 0.02 difference in the persistence probabilities, with no uncertainty analysis, and the asymptotic criterion is stated incorrectly. These are fixable with additional analysis and discussion, so I recommend major revision. I would also flag to the editor that the source parameter estimates in [17] may themselves carry uncertainty that is not propagated here."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: this is a solid, careful theory paper that does something genuinely new. It couples a telegraph-with-pauses model of single-dendrite mRNA motion to a semi-Markov chain on reconstructed dendritic trees, and it derives two asymptotic regimes. The math looks sound; the single-dendrite absorption probabilities and conditional MFPTs are derived in the supplement, and the Monte Carlo agreement in Fig. 3 is reassuring. The code is on GitHub.\n\nWhat is actually new is the qualitative regime separation: when retrograde persistence dominates (β > α), the soma-to-synapse CMFPT is asymptotically flat across the arbor; when anterograde dominates, it grows linearly with path distance. That is a compact mechanistic prediction not present in the cited PTPP or intracellular-transport literature. The semi-Markov framework is standard, but applying it to full reconstructed morphologies is a legitimate step.\n\nNow the soft spots. First, the assignment of real neurons to the retrograde regime rests on α = 0.51, β = 0.53 from Song et al. [17] — a difference of 0.02 with no quoted uncertainty. That is load-bearing: flip the sign and the flat regime becomes linear, or vice versa. Unless the persistence bias is measured with error bars, the real-neuron claims in Fig. 4 are illustrative rather than evidential. The stress-test note also correctly flags that the asymptotic criterion should be l ≫ v/(λ|β − α|), not just l ≫ v/λ. For β − α = 0.02 the crossover is about 11 µm, so the long dendrites are fine, but if the true difference were 0.001, the crossover would be ~230 µm and the asymptotics would not apply. The paper should state the criterion correctly.\n\nSecond, the retrograde-regime CMFPTs are enormous — ~6×10^7 s for the human neuron — orders of magnitude longer than typical mRNA half-life. The author acknowledges degradation as future work, but that is not a minor caveat when the \"equal age\" interpretation is the headline. Either add a degradation term or restrict the claim to biologically relevant time scales.\n\nThe simplifications (uniform branch choice, length-only morphology) are acknowledged; they are acceptable for a first model, though they limit quantitative matching.\n\nWho this is for: people working on stochastic transport on filamentous networks or on mRNA localization. It deserves serious refereeing — the theory is coherent, verifiable, and honest about many of its limitations. I would want the β − α sensitivity addressed and the asymptotic criterion fixed before acceptance, but this is not a desk reject.\n\nFor peer review: yes, send it out, with a request for careful revision on the points above.","headline":"A careful two-scale stochastic model with a genuinely new regime prediction, but the real-neuron claims hinge on an unquantified 0.02 persistence bias.","tokens_in":22846,"tokens_out":3621,"would_cite":true,"duration_ms":32304,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60K15","60J27","92C20"],"pacs":[],"model":"deepseek-v4-flash","headline":"A 0.02 direction-bias difference decides how mRNA arrival times spread across synapses.","keywords":["mRNA transport","persistent telegraph process with pauses","semi-Markov model","mean first-passage time","dendritic tree","microtubule motor transport","stochastic transport in neurons"],"falsifier":"Measure $\\alpha$ and $\\beta$ in the same neuron types with enough single-molecule trajectories to make the 0.02 difference statistically significant; if the confidence interval covers zero or the sign reverses, the retrograde-bias prediction for human Purkinje and mouse pyramidal cells fails. Alternatively, track individual mRNA cargoes in a reconstructed long-dendrite neuron: near-equal conditional mean first-passage times at near and far synapses under $\\beta>\\alpha$ versus linear growth with distance under $\\alpha>\\beta$ are mutually exclusive signatures.","tokens_in":21751,"feed_emoji":"🧬","tokens_out":9609,"duration_ms":80323,"temperature":0.7,"pith_summary":"The paper builds a two-level stochastic model for mRNA cargo moving along microtubules from the soma to synapses. At the single-dendrite level, motion is a persistent telegraph process with pauses: runs at constant speed in either direction, exponential run and rest durations, and a tendency to keep the same direction after a pause. These single-dendrite results are embedded in a semi-Markov walk over the dendritic tree, whose nodes are the soma, branching points, and synapses. The central prediction is a regime switch: when retrograde persistence is stronger ($\\beta>\\alpha$), the conditional mean first-passage time to every synapse is nearly the same regardless of distance, while when anterograde persistence is stronger ($\\alpha>\\beta$), travel time grows nearly linearly with soma-synapse distance, as if the whole tree were one dendrite. Numerical case studies assign real neurons to the retrograde regime using the measured values $\\alpha=0.51$, $\\beta=0.53$.","feed_headline":"A 2 percent direction bias decides how mRNA reaches synapses","feed_subtitle":"Slightly stronger retrograde persistence equalizes travel times; anterograde bias makes them grow linearly with distance.","key_machinery":"The central object is the Persistent Telegraph Process with Pauses (PTPP), a continuous-time Markov process with four states: running and resting in the anterograde and retrograde directions. Runs end at rate $\\lambda$, rests at rate $\\mu$; after a rest, the cargo continues in the same direction with probability $\\alpha$ (after an anterograde run) or $\\beta$ (after a retrograde run), and reverses with the complementary probability. For a dendrite of length $l$, the paper solves the infinitesimal generator equations for the probability $p^l(0)$ of traversing from the soma-side origin to the far end and for the weighted MFPT $w^l(0)$, then uses the conditional MFPT $c^l(0)=w^l(0)/p^l(0)$. These single-dendrite quantities become the transition probabilities and sojourn times of a semi-Markov model whose states are the soma, branching nodes, and synapses and whose edges are dendrites, with a uniform $1/d_i$ choice probability at each branching node. The fundamental matrix of the embedded absorbing Markov chain then yields absorption probabilities and conditional MFPTs to each synapse, and asymptotic analysis for $l\\gg v/\\lambda$ produces the two-regime formulas.","core_discovery":"The central claim is that for dendrites much longer than the typical run distance $v/\\lambda$, transport in the whole dendritic tree is governed by the sign of the persistence bias. With $\\beta>\\alpha$, the probability of traversing a dendrite in the anterograde direction is exponentially small, the cargo spends a long quasi-stationary period wandering among transient nodes, and the conditional mean first-passage time (CMFPT) to each synapse converges to a common value independent of soma-synapse distance. With $\\alpha>\\beta$, the probability of reaching the far end of a dendrite in the anterograde direction saturates to a nonzero constant, return probabilities and loop times become nearly constant, and the soma-to-synapse CMFPT becomes a sum of per-dendrite traversal times, hence asymptotically linear in distance with slope $S_{\\mathrm{ant}}=[(1+\\rho)(2-\\alpha-\\beta)]/[v(\\alpha-\\beta)]$; in the $\\beta>\\alpha$ case the slope is $S_{\\mathrm{ret}}=[(1+\\rho)(2-\\alpha-\\beta)-\\rho(\\beta-\\alpha)^2]/[v(\\beta-\\alpha)]$. The symmetric case $\\alpha=\\beta$ is qualitatively different, producing quadratic growth in dendrite length, the footprint of diffusive behavior.","pith_inferences":["Editorial extension: with only a 0.02 difference between $\\alpha$ and $\\beta$, the predicted regime for real neurons is fragile; re-estimating the two probabilities with confidence intervals narrower than 0.02 would either confirm or eliminate the retrograde-regime assignment.","Editorial extension: the model omits mRNA degradation; adding a finite half-life would make far synapses receive fewer, not just younger, mRNA molecules, turning the near-uniform arrival-time prediction into a distance-dependent survival correction.","Editorial extension: replacing the uniform branch-choice probability $1/d_i$ with widths or microtubule-number-dependent choices could shift the effective threshold between the two regimes, although the asymptotic structure of the two regimes should survive.","Editorial extension: the near-constant arrival time under $\\beta>\\alpha$ acts like an age filter on synaptic mRNA; if confirmed experimentally, it would imply that differences in mRNA age are not a mechanism for synapse-to-synapse plasticity differences in that regime."],"forward_implications":["For $\\beta>\\alpha$ and dendrites long relative to $v/\\lambda$, the model predicts that mRNA molecules arriving at different synapses have nearly identical ages, so synapse-to-synapse differences in arrival time essentially disappear.","For $\\alpha>\\beta$, the soma-to-synapse CMFPT becomes a sum of traversal times along dendrites on the path, so it grows linearly with distance with slope $S_{\\mathrm{ant}}=[(1+\\rho)(2-\\alpha-\\beta)]/[v(\\alpha-\\beta)]$, as if transport happened on a single long dendrite.","In the symmetric case $\\alpha=\\beta$, the single-dendrite CMFPT grows quadratically in dendrite length, so the two biased regimes are distinguished from the diffusive limit by the scaling exponent of length (1 versus 2).","Under retrograde bias, the slope of the linear term can be smaller than under anterograde bias, and for some parameter regions the average travel time in the direction of smaller persistence is actually shorter, so stronger persistence need not mean faster transport."],"supporting_citations":[{"why":"Supplies the empirical run-and-pause parameter estimates, including $\\alpha=0.51$ and $\\beta=0.53$, used to place real neurons in the retrograde-bias regime.","marker":"[17]"},{"why":"Supplies the reconstructed neuronal morphologies used for the numerical CMFPT calculations.","marker":"[31]"},{"why":"Provides the human Purkinje-cell morphology whose calculations show nearly constant CMFPT under retrograde bias.","marker":"[38]"},{"why":"Provides the mouse pyramidal-neuron morphology whose calculations show linear CMFPT dependence under anterograde bias.","marker":"[39]"},{"why":"Introduces the persistent telegraph process with pauses that the paper extends to the single-dendrite level.","marker":"[18]"},{"why":"Provides the absorbing Markov chain theory, including the fundamental matrix, used to derive the semi-Markov WMFPT formulas.","marker":"[35]"},{"why":"Contains the derivations of the generator equations, the asymptotic slopes, and the semi-Markov matrix formulas used throughout.","marker":"[33]"}],"fun_headline_variants":["A small direction bias dictates mRNA travel in neurons","Retrograde bias equalizes mRNA delivery; anterograde makes it distance-based","mRNA transport: bias direction decides synaptic arrival times","Subtle persistence bias shapes mRNA's journey to synapses","Neuron mRNA: how a 2% bias sets travel times"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the measured retrograde persistence probability $\\beta=0.53$ really exceeds the anterograde one $\\alpha=0.51$; the values differ by only 0.02 with no quoted uncertainty, and the real-neuron regime flips if the true sign is reversed, with the asymptotic slopes diverging as $\\beta-\\alpha\\to 0$.","fun_headline_variants_meta":{"raw":{"variants":["A small direction bias dictates mRNA travel in neurons","Retrograde bias equalizes mRNA delivery; anterograde makes it distance-based","mRNA transport: bias direction decides synaptic arrival times","Subtle persistence bias shapes mRNA's journey to synapses","Neuron mRNA: how a 2% bias sets travel times"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000216,"raw_usage":{"total_tokens":1481,"prompt_tokens":1044,"completion_tokens":437,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":660,"completion_tokens_details":{"reasoning_tokens":353}},"tokens_in":660,"tokens_out":437,"duration_ms":3862,"temperature":1.0,"reasoning_tokens":353,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:15:07.925630+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure $\\alpha$ and $\\beta$ in the same neuron types with enough single-molecule trajectories to make the 0.02 difference statistically significant; if the confidence interval covers zero or the sign reverses, the retrograde-bias prediction for human Purkinje and mouse pyramidal cells fails. Alternatively, track individual mRNA cargoes in a reconstructed long-dendrite neuron: near-equal conditional mean first-passage times at near and far synapses under $\\beta>\\alpha$ versus linear growth with distance under $\\alpha>\\beta$ are mutually exclusive signatures.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the empirical run-and-pause parameter estimates, including $\\alpha=0.51$ and $\\beta=0.53$, used to place real neurons in the retrograde-bias regime."},{"cited_title":"L´ evy, inProc","cited_arxiv_id":null,"evidence_quote":"Supplies the reconstructed neuronal morphologies used for the numerical CMFPT calculations."},{"cited_title":"Amrute-Nayak and S","cited_arxiv_id":null,"evidence_quote":"Provides the human Purkinje-cell morphology whose calculations show nearly constant CMFPT under retrograde bias."},{"cited_title":"Masoli, D","cited_arxiv_id":null,"evidence_quote":"Provides the mouse pyramidal-neuron morphology whose calculations show linear CMFPT dependence under anterograde bias."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the persistent telegraph process with pauses that the paper extends to the single-dendrite level."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the absorbing Markov chain theory, including the fundamental matrix, used to derive the semi-Markov WMFPT formulas."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Contains the derivations of the generator equations, the asymptotic slopes, and the semi-Markov matrix formulas used throughout."}],"review_version":1}