{"id":"34484468-df79-4f37-af11-69d94c7b61e2","arxiv_id":"2411.14588","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A two-type ecDNA model shows switching does not alter total copy number distribution under equal fitness, but allows mixed cells to persist.","lead":"This paper presents a mathematical model for cells carrying two different types of extrachromosomal DNA (ecDNA) that can switch between types. It proves that under equal fitness switching does not affect the total ecDNA copy number distribution, while enabling cells carrying both types to persist.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claim that switching maintains mix cells depends on the untested assumption that mix cells pay no fitness cost; even a small cost could invalidate the persistence/dominance result and the switching-independence of total load.","rationale":"The reader's weakest-assumption analysis identifies the same load-bearing concern: the mix-cell fitness assumption. My independent reading confirms that this assumption is what makes the exact reduction to a single-type model possible under identical selection, and it is also the basis for the mix-cell persistence/dominance conclusion. The paper cites only an unpublished preprint for this assumption and explicitly says it can be relaxed, but does not explore the consequences. The alternative candidate concern, the uncontrolled truncation of infinite moment sums (Eq. 24), is real but less central because the paper validates its approximate moment equations against stochastic simulations of the same model; the mix-cell fitness assumption affects the model itself, not just the approximation scheme. A targeted computational perturbation of smix would settle whether the headline biological claim survives a plausible cost for carrying both ecDNA types. Since this concern does not change the reader's CONDITIONAL verdict, no verdict adjustment is needed.","tokens_in":26536,"tokens_out":10513,"duration_ms":108219,"concrete_test":"Rerun the model's Gillespie simulations with mix-cell fitness smix = max(sy, sr) - c for c in {0.01, 0.05, 0.1, 0.2} (or equivalently smix = max(sy, sr)/(1+c)), over the same parameter sweeps as the paper (py = pr = p in {0, 0.01, 0.1, 0.5}; sy = sr in {1, 2}; sy = 2.5, sr = 1). Track the asymptotic mix-cell fraction among ecDNA-positive cells and the total copy-number distribution. If for any small c > 0 mix cells no longer persist or dominate, or the total distribution develops a dependence on p, the no-cost assumption is the load-bearing condition and the headline claim must be qualified. If the qualitative behavior is unchanged for all c tested, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's novel persistence/dominance result rests on the assumption, stated in Methods and Figure 1d, that a cell carrying both ecDNA types has reproduction rate max(sy, sr), i.e. no extra fitness burden from carrying both types. This assumption is load-bearing in two ways. First, when sy = sr = s, it is what makes every ecDNA-positive cell have fitness s, so the exact reduction to the single-type equation (Eq. 9) goes through; if mix cells paid a cost c > 0, the total copy-number dynamics would depend on the mix fraction and hence on the switching rates, invalidating the headline 'switching independence' for total ecDNA load. Second, the claim that mix cells can persist and even dominate under switching (Discussion) is demonstrated only under this no-cost assumption; a fitness cost would change the mix-cell terms in the master equations (Eqs. 5-8) and could remove the maintenance/dominance effect. The assumption is supported only by a citation to ref. [28], an unpublished preprint, and the text explicitly says it 'can be easily released' without exploring consequences. Since the central biological message is that switching can maintain multiple ecDNA types, this parameter assumption is the most load-bearing unverified input. The mathematical derivation itself (Eqs. 9-14) appears correct conditional on the model assumptions, so the concern is not internal inconsistency but model robustness.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper develops a master-equation framework for cell populations carrying two types of extrachromosomal DNA (ecDNA), with independent binomial segregation, possible switching between types at cell division, and selection. The model covers two ecDNA 'species' (no switching), two genotypes/phenotypes (one-way or two-way switching), and the resulting subpopulations of pure yellow, pure red, mix, and ecDNA-free cells. The central exact result is that when the two ecDNA types have identical fitness, the total number N_k of cells with k ecDNA copies satisfies the single-type equation -sN_k + 2s Σ_{i≥⌈k/2⌉} N_i (2i choose k)/2^{2i}, independent of the switching rates (Eq. 9); consequently the total ecDNA-positive distribution and the fraction of ecDNA-free cells are claimed to be switching-independent. For the subpopulations, the authors derive approximate first- and second-moment ODEs using a four-term truncation of infinite sums and small-p Taylor expansions, and compare them with Gillespie simulations. They conclude that without switching, mix cells cannot be maintained, while with switching they can persist and even dominate the ecDNA-positive population. The model assumes that mix cells have the fitness of the fitter pure type, i.e., no extra fitness cost for carrying both ecDNA types.","tokens_in":26792,"tokens_out":16766,"duration_ms":150734,"significance":"If Eq. (9) and the supporting analyses are correct, the paper provides a useful reduction result: selectively equivalent ecDNA types do not affect the total ecDNA load regardless of switching, while switching can maintain multiple ecDNA types in a way that non-switching species cannot. The combinatorial proof of Eq. (9) is a genuine strength, as is the systematic comparison with Gillespie simulations in Figures 3-6 and the clear separation of exact from approximate results. The main caveats are that the closed form in Eq. (18) relies on an undefined quantity, the moment approximations use an uncontrolled truncation, and the persistence/dominance conclusions depend on an untested no-cost assumption for mix cells. With these points resolved, the framework would be a valuable contribution to the quantitative cancer-evolution literature.","major_comments":[{"comment":"The function δ(s) is listed in Table 1 as a 'zeroing function' but is never defined or derived, so the claimed closed form ρ0,0(t)=t/(2+t e^{δ(s)t}) for sy=sr>1 cannot be verified. In particular, the statement that ρ0,0 shrinks to zero requires a sign and magnitude condition on δ(s) that is not given. Please provide the definition and derivation, or label Eq. (18) as an ansatz and support it with simulations.","section":"Under neutral and identical positive selection, Eq. (18)"},{"comment":"The equation for C0,0 as printed includes the term 2sr Σ_{h=0}∞ C0,h (2h choose 0)/2^{2h}; with h=0 this contributes 2sr C0,0 on top of the -C0,0+2C0,0 division term already present. For sr=1 the free-cell compartment would grow at three times its correct rate. The lower index should be h≥1, and the first sum in that line should be clarified, since as printed 'j+h=1' is ambiguous. Please correct the complete system.","section":"Methods, Eq. (8), last line"},{"comment":"The infinite sums E∞_j and V∞_j are truncated to their first four terms with only the heuristic statement that the terms decay geometrically in h. No bound on the remainder is given, and these truncated moment equations are part of the evidence for the persistence/dominance of mix cells under switching. Please add a quantitative truncation-error estimate or a convergence check by increasing the truncation order, or explicitly state that the persistence claim rests on the Gillespie simulations rather than on the truncated moment equations.","section":"Moment analysis for multiple ecDNA species when switching is off, Eq. (24) and Tables 2-5"},{"comment":"The Total row gives dM^(1)/dt=0 for sy=sr>1, which contradicts Eq. (27) and Table 2, where dM^(1)/dt=(s-1)ρ0,0M^(1). Since Figure 5 compares subpopulation moments against this system, the contradiction should be resolved and the normalization of the plotted moments clarified.","section":"Under identical fitness and one-way switching, Table 4"},{"comment":"The assumption that mix cells have reproduction rate max(sy,sr), i.e., no extra fitness cost or benefit from carrying both types, is load-bearing for the exact reduction in Eq. (9) when sy=sr and for the conclusion that mix cells can persist and dominate under switching. The text notes this assumption 'can be easily released' but does not explore the consequences, and ref. [28] is an unpublished preprint. Please provide a sensitivity analysis with a mix-cell fitness cost (e.g., max(sy,sr)-c in Eqs. (5)-(8)) and state whether the switching-independence and persistence results survive, or explicitly limit the conclusions to the no-cost regime.","section":"Methods, 'A general framework of two ecDNA types' and Figure 1d"}],"minor_comments":[{"comment":"The rates γ(t), g1, g2, h1, h2, k1, k2 in Eq. (2) are listed in Table 1 but never defined; please define them or remove Eq. (2) if it is superseded by Eqs. (8).","section":"Methods, Eq. (2)"},{"comment":"Eq. (36) has '+ (s-1)ρ0,0(t)ρi,0(t)' but Eq. (37) and Table 4 use the opposite sign in the corresponding moment equation; please align the signs.","section":"Eq. (36) versus Eq. (37) and Table 4"},{"comment":"The notation ρ0(τ) in Eq. (17) should presumably be ρ0,0(τ), the frequency of ecDNA-free cells; please use one consistent notation.","section":"Eq. (17)"},{"comment":"The sentence refers to Figure 2b&d for intermediate switching values, but those panels show the non-identical fitness scenario; please rephrase to distinguish identical switching rates (py=pr) from identical selection (sy=sr).","section":"Discussion, paragraph beginning 'Moreover, if the switching is identical and two-way'"},{"comment":"References [30] and [31] are duplicates, and reference [10] cites the current manuscript as 'in preparation'; please remove the self-citation or replace it with an appropriate published reference.","section":"References"},{"comment":"The x-axis labels are inconsistent across panels (number of cells vs. time in generations); please unify the axis convention and state clearly how the number of cells relates to generation time.","section":"Figures 3-5"},{"comment":"The statement that the term in Eq. (13) '= 1' should be stated more carefully: the sum over k of the convolution is 1 for each v, and the additional summation over v yields the binomial identity in Eq. (14).","section":"Eqs. (13)-(14)"}],"recommendation":"major_revision","confidential_remarks":"The paper's core exact reduction (Eq. 9) appears correct and is a useful contribution, and the simulation comparisons are well matched to the analytical approximations. The main risks are fixable: the undefined δ(s) in Eq. (18), the apparent error in the C0,0 equation, the inconsistent total-moment row in Table 4, and the unexamined mix-cell fitness assumption. I do not see grounds for rejection if these points are addressed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe thing you should know: this paper contains one genuinely rigorous result and one load-bearing assumption that is never stress-tested. The rigorous result is that when the two ecDNA types have equal fitness, the total copy-number distribution is independent of switching and reduces exactly to the single-type equation (Eq. 9). The proof via summing binomial identities is correct, and the simplification to Eq. (9) checks out. That is a real contribution. The moment analysis for pure, mix, and free subpopulations is also new, and the comparisons with Gillespie simulations are honest.\n\nWhere it does well: the framework cleanly unifies three biological scenarios (species, genotypes, phenotypes) by changing one parameter. The paper is clearly written and the math is mostly careful. The authors state their approximations, and the small-MSE matches for small switching probabilities give some confidence.\n\nThe soft spots are real but not fatal. Eq. (18) introduces an undefined δ(s); that formula cannot be checked as written. The moment equations truncate infinite sums to four terms without an error estimate. The authors validate only against simulations of the same model, so the biology rests entirely on assumptions. The biggest one: mix cells are assumed to have fitness max(sy, sr), i.e. no extra cost or benefit from carrying both types. That assumption is cited to an unpublished preprint, and the paper itself says it 'can be easily released' but never does it. If mix cells paid even a small cost, the switching-independence theorem would likely break, because the master equations for different subpopulations would no longer carry the same fitness factor, and the persistence/dominance claim for mix cells could vanish. This is the key unexamined robustness question.\n\nWho this is for: mathematical oncology and anyone modeling ecDNA dynamics. A careful reader gets a solid two-type framework and a clear statement of where the biology is guesswork.\n\nRecommendation: this deserves a serious referee. The core derivation is correct and new; the paper needs revision to define δ(s), bound the truncation error, and at least discuss what happens when mix cells carry a fitness cost. I would engage with it.","headline":"Rigorous two-type ecDNA result with an unexamined mix-fitness assumption that deserves referee time.","tokens_in":27320,"tokens_out":2702,"would_cite":true,"duration_ms":26697,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["92D25","60J85"],"pacs":[],"model":"deepseek-v4-flash","headline":"Under equal fitness, switching among ecDNA types does not change total copy-number dynamics.","keywords":["extrachromosomal DNA (ecDNA)","random segregation","type switching","copy number distribution","branching process","intratumor heterogeneity","mathematical oncology","population dynamics"],"falsifier":"Simulate Eq. (8) with mix-cell fitness set to $\\max(s_y,s_r)-\\varepsilon$ for a range of positive $\\varepsilon$ and compare the mixed-cell fraction over time; if even a small fitness cost removes the predicted persistence and dominance, the core persistence claim fails.","tokens_in":26314,"feed_emoji":"🧬","tokens_out":5676,"duration_ms":56851,"temperature":0.7,"pith_summary":"This paper builds a mathematical model of tumor cells carrying two types of extrachromosomal DNA (ecDNA), each with its own fitness and its own probability of switching into the other type. Its central result is that when the two types have equal selection strength, the total number of ecDNA copies per cell evolves exactly as in the known single-type model, independent of switching rates. The fraction of ecDNA-free cells is likewise unaffected by switching. Switching only reshuffles cells between pure-type and mixed subpopulations; it does not change the overall ecDNA burden. The paper also shows that without switching, mixed cells cannot be maintained, whereas two-way switching lets them persist and even dominate the ecDNA-positive population, which matters for tumor heterogeneity and resistance.","feed_headline":"Equal-fitness ecDNA types: switching never changes total copy load","feed_subtitle":"Two-type model collapses to single-type equation; only unequal selection makes switching matter.","key_machinery":"The central object is a master equation counting cells by copy numbers of two ecDNA colors, $C_{i,k}(t)$. The load-bearing identity is that the sum over all switching outcomes of a binomial distribution, or of a convolution of two binomials, equals one, because switching only relabels copies without changing how many copies are inherited. Summing the two-color equations over color partitions collapses the process to the single-type branching process of Eq. (9). The approximate moment equations then track weighted first and second moments of pure and mixed subpopulations, with an infinite geometric-like sum truncated to its first four terms.","core_discovery":"The discovery is a reduction: by summing the master equation over all color configurations, all switching terms cancel via complete binomial probability sums, leaving $\\frac{dN_k}{dt} = -sN_k + 2s \\sum_{i \\ge \\lceil k/2 \\rceil} N_i \\binom{2i}{k} 2^{-2i}$, which is exactly the single-type ecDNA equation. Therefore, for $s_y = s_r = s \\ge 1$, the distribution of total ecDNA copies and the fraction of ecDNA-free cells are independent of $p_y$ and $p_r$ at all times. With $s_y \\neq s_r$ the cancellation fails and switching shapes total copy dynamics. For ecDNA species without switching, mixed-type cells are transient; with two-way switching, mixed-type cells can persist and dominate the ecDNA-positive population.","pith_inferences":["Extension: if mix cells pay any fitness cost, the persistence and dominance of mixed cells under switching is fragile; rerunning Eq. (8) with mix fitness set to $\\max(s_y,s_r)-\\varepsilon$ would show how quickly the effect disappears.","Unstated consequence: for equal fitness, therapies aimed at lowering total ecDNA copy number cannot be tuned by knowing switching rates; only the sum of copies matters, while type composition remains relevant for heterogeneity and resistance.","Testable extension: the four-term truncation of the moment sums can be stress-tested by comparing the approximate moment equations against the untruncated master equation for larger $p$ or higher moments, to see whether the predicted mix-cell dominance is quantitatively robust."],"forward_implications":["For cells carrying two ecDNA types with equal fitness, the total ecDNA copy distribution and the fraction of ecDNA-free cells can be predicted from the single-type model without measuring switching rates.","When selection is identical, switching affects only the composition of cells among pure-yellow, pure-red, and mixed subpopulations, not the overall ecDNA burden.","Under two-way switching, mixed cells can persist and even dominate the ecDNA-positive population despite receiving no extra fitness benefit, with intermediate switching rates most effective.","If the two ecDNA types have different fitness, switching changes the total ecDNA dynamics, so type identity cannot be ignored.","The framework generalizes to more than two ecDNA types, allowing the same switching parameter to model species, genotypes, or phenotypes."],"supporting_citations":[{"why":"Supplies the single-type ecDNA branching model to which the equal-fitness two-type dynamics reduce, including its experimental and clinical validation.","marker":"[22]"},{"why":"Provides the experimental basis for multiple ecDNA species coexisting in one cell and for assigning mix cells the fitness of the fitter pure type.","marker":"[28]"},{"why":"Documents uneven or random segregation of ecDNA at cell division, motivating the independent binomial partition rule used throughout the model.","marker":"[11]"},{"why":"Earlier stochastic treatment of binomial partitioning of cellular components, which the paper notes reduces to its framework when replication outside division is excluded.","marker":"[18]"},{"why":"Provides closed-form stochastic solutions for non-equilibrium inheritance over many divisions, used as a consistency check for the single-type reduction.","marker":"[19]"}],"fun_headline_variants":["Equal fitness makes ecDNA switching irrelevant to total copy numbers","Without switching, mixed ecDNA cells vanish; with switching, they persist","When ecDNA fitness differs, switching drives mixed-type dominance","Equality cancels switching terms: total ecDNA copy count unchanged","Two-type ecDNA model collapses to one when fitness is equal"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The conclusion that mixed cells can persist under switching rests on treating a cell that carries both ecDNA types as no fitter and no sicker than a cell carrying only the fitter type, an assumption the paper adopts without testing alternatives.","fun_headline_variants_meta":{"raw":{"variants":["Equal fitness makes ecDNA switching irrelevant to total copy numbers","Without switching, mixed ecDNA cells vanish; with switching, they persist","When ecDNA fitness differs, switching drives mixed-type dominance","Equality cancels switching terms: total ecDNA copy count unchanged","Two-type ecDNA model collapses to one when fitness is equal"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000629,"raw_usage":{"total_tokens":2905,"prompt_tokens":940,"completion_tokens":1965,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":556,"completion_tokens_details":{"reasoning_tokens":1878}},"tokens_in":556,"tokens_out":1965,"duration_ms":14010,"temperature":1.0,"reasoning_tokens":1878,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:07:18.660054+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate Eq. (8) with mix-cell fitness set to $\\max(s_y,s_r)-\\varepsilon$ for a range of positive $\\varepsilon$ and compare the mixed-cell fraction over time; if even a small fitness cost removes the predicted persistence and dominance, the core persistence claim fails.","supporting_citations":[{"cited_title":"The evolutionary dynamics of extrachromosomal DNA in human cancers","cited_arxiv_id":null,"evidence_quote":"Supplies the single-type ecDNA branching model to which the equal-fitness two-type dynamics reduce, including its experimental and clinical validation."},{"cited_title":"Coordinated inheritance of extrachromosomal DNA species in human cancer cells.bioRxiv, 2023","cited_arxiv_id":null,"evidence_quote":"Provides the experimental basis for multiple ecDNA species coexisting in one cell and for assigning mix cells the fitness of the fitter pure type."},{"cited_title":"Live- cell imaging shows uneven segregation of extrachromosomal DNA elements and transcriptionally active extrachromosomal DNA hubs in cancer.Cancer Discov- ery, 12:468–483, 2022","cited_arxiv_id":null,"evidence_quote":"Documents uneven or random segregation of ecDNA at cell division, motivating the independent binomial partition rule used throughout the model."},{"cited_title":"Stochastic modelling, bayesian inference, and new in vivo mea- surements elucidate the debated mtDNA bottleneck mechanism.Elife, 4:e07464, 2015","cited_arxiv_id":null,"evidence_quote":"Earlier stochastic treatment of binomial partitioning of cellular components, which the paper notes reduces to its framework when replication outside division is excluded."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides closed-form stochastic solutions for non-equilibrium inheritance over many divisions, used as a consistency check for the single-type reduction."}],"review_version":1}