{"id":"a1cb721c-3149-4704-b56a-d1cb8c8e6ab1","arxiv_id":"2412.12418","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"The entropy and enthalpy components of excitation energy transfer in the PSII supercomplex are computed from a structure-based rate matrix, revealing subunit-specific roles and a tunable entropic design principle.","lead":"Using a structure-based model of the Photosystem II supercomplex, the authors show that the entropy of the excitation distribution changes over time in ways that differ between protein subunits. They argue that this entropic variability lets each subunit play a distinct role, enabling efficient trapping, photoprotection, and robustness when one reaction center is closed.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Population contraction times and entropy curves are computed from a single disorder-averaged rate matrix (e^{⟨K⟩t}) rather than by averaging dynamics over realizations (⟨e^{K(δ)t}⟩), so the reported subunit-specific entropy patterns may be artifacts of that averaging; no error bars are shown.","rationale":"I read the paper as making a specific, testable modeling claim: that after a defined initial excitation, the Shannon entropy of the exciton population distribution rises and then falls, and the time of the maximum ('population contraction time') is a subunit-specific design feature that can be tuned by LHCII binding. For that claim to hold, the computed contraction times and entropy curves must be reliable outputs of the stochastic model. The weakest point in that chain is not the philosophical status of Shannon entropy (though the detailed-balance caveat is real) but the numerical treatment of static disorder: the authors generate 500 disorder realizations and then average the rate matrices before propagating the population. Averaging rates and exponentiating is not the same as exponentiating each rate matrix and averaging the populations, and the nonlinearity of both the rates and the entropy functional means the published numbers can be biased in an unknown direction. The absence of error bars is not merely a presentational gap; it is a direct symptom of this averaging choice. The proposed test is feasible because the rate-matrix construction code is public. If the per-realization analysis reproduces the ordering and the averaged-matrix values lie within the spread, the conditional verdict can stand. If not, the central claim is unsupported as written. I agree partially with the reader: the detailed-balance/quantum-dynamics concern is related, but the disorder-averaging issue is more immediately decisive for the numbers.","tokens_in":11507,"tokens_out":8388,"duration_ms":84599,"concrete_test":"Rerun the full pipeline on each of the 500 disorder realizations: for realization δ, build K_δ, evolve P_δ(t)=e^{K_δ t}P(0), compute ΔS_δ(t) and the population contraction time (argmax ΔS_δ), and also compute the average entropy curve ⟨ΔS_δ(t)⟩_δ and the distribution of contraction times. Compare these to the values in Figs. 2–5 obtained from e^{⟨K⟩t}. If the published contraction times fall outside the central 68% interval of the per-realization distribution, or if the subunit ordering (CP43 < CP26 < CP47 < CP29 < S-LHCII < M-LHCII) changes, the central entropy-design claim is not supported by the current analysis.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative results—the population contraction times and the entropy/enthalpy curves in Figs. 2–5—are all computed from a single averaged rate matrix. Methods states: '500 sets of site energies are generated to construct 500 rate matrices. The calculation of entropy is based on the averaged rate matrix from these 500 realizations.' This replaces the correct disorder-averaged dynamics ⟨e^{K(δ)t}P(0)⟩_δ with e^{⟨K⟩_δ t}P(0), and then evaluates the entropy as S(⟨p(t)⟩_δ) rather than ⟨S(p_δ(t))⟩_δ. Rate constants depend nonlinearly (often exponentially) on the random site-energy offsets, so the average rate matrix is not the rate matrix of a representative complex; and because Shannon entropy is concave, Jensen's inequality guarantees S(⟨p⟩) ≥ ⟨S(p)⟩, so the entropy curves are biased upward by construction. The reported contraction times are therefore not necessarily the contraction times of any typical realization or of the ensemble. Since the design-role conclusions are built on comparisons of these times across subunits and across C2S2M2 vs C2S2, a reanalysis with explicit disorder-averaged populations is required before the entropy-as-design-principle claim can be considered supported. The manuscript gives no distribution across the 500 realizations for any reported time, despite having generated them.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs structure-based rate matrices for the C2S2M2 and C2S2 Photosystem II supercomplexes, solves the resulting master equation for excitation population dynamics, and computes time-dependent changes in Shannon entropy and enthalpy for many initial excitation conditions. From these calculations the authors define a population contraction time and argue that different protein subunits play distinct functional roles: D1-side subunits direct energy toward the reaction center, D2-side subunits support bidirectional inter-monomer energy flow, and LHCII binding tunes the entropic character of the energy landscape. The central claims are that entropy is a design principle in PSII and that subunit-specific entropy/enthalpy balances enable both efficient trapping and photoprotection.","tokens_in":11761,"tokens_out":4197,"duration_ms":43572,"significance":"If the calculations are correct, the paper would provide a useful conceptual framework for interpreting PSII's flat energy landscape and would connect stochastic-thermodynamic quantities to structural biology. The strengths of the manuscript include its use of high-resolution structures, explicit comparison of C2S2M2 and C2S2 complexes, treatment of open and closed reaction centers, and the public availability of the rate-matrix construction code. However, the central quantitative results are computed from a single disorder-averaged rate matrix rather than from ensemble-averaged dynamics, and the interpretation of the Shannon entropy as a thermodynamic entropy is not fully justified. These issues affect the load-bearing comparisons across subunits and between complexes, so the design-principle conclusions are not yet supported in their present form.","major_comments":[{"comment":"The calculation of entropy uses the Shannon entropy of the exciton population distribution, S = -k_B sum_i p_i ln p_i, and the paper interprets this as the thermodynamic entropy in the free-energy decomposition of Eq. (2). The statement that the rates satisfy detailed balance with respect to a Boltzmann distribution is a consistency condition on the rate matrix, but it does not by itself establish that this population Shannon entropy equals the thermodynamic entropy of the exciton-phonon system; coherences, bath entropy, and non-Markovian effects are omitted. Since the paper's title and main conclusions are about entropy as a thermodynamic design principle, this identification needs either a derivation or an explicit statement that the quantity is a configurational entropy of the population distribution whose thermodynamic interpretation is an assumption.","section":"II.A and Eq. (1)"},{"comment":"The classification of subunits into D1-side directed trappers, D2-side bidirectional spreaders, and LHCII-mediated inter-monomer connectors is based on site-resolved population contraction times and same-monomer trapping probabilities computed from the single averaged rate matrix. Because disorder affects connectivity and trapping probabilities nonlinearly, these classifications could change under proper ensemble averaging. The paper should report per-realization statistics for contraction times and trapping probabilities, and the qualitative design conclusions should be shown to be robust to the disorder-averaging procedure before being presented as general design principles.","section":"III, Figs. 3-5 and Section II.A"}],"minor_comments":[{"comment":"The beginning of the Concluding Remarks contains the typo 'rIn' instead of 'In'.","section":"IV"},{"comment":"The caption says 'In order of A-E, excitations are localized in...' but the figure contains panels A-H; the text should say A-H.","section":"Fig. 2 caption"},{"comment":"The phrase 'Chls a 610 and 612 of the LHCII (B) complexes' should be 'Chl a 610 and Chl a 612' for grammatical consistency.","section":"III"},{"comment":"The text calls an R^2 value of 0.68 'not a very strongly correlated relationship'; this wording is surprising because 0.68 is usually considered a moderate-to-strong correlation, and the conclusion would be clearer if the authors stated the quantitative threshold they use.","section":"III, Fig. S5"},{"comment":"The notation C2S2M2 appears with inconsistent spacing (e.g., 'C 2S2M2' in several places); please standardize throughout.","section":"II.A"}],"recommendation":"major_revision","confidential_remarks":"The disorder-averaging issue is the main technical obstacle and is fixable within the manuscript's scope by recomputing the dynamics with ensemble averaging over the 500 realizations and reporting distributions. I did not see evidence of novelty or attribution problems; the paper's contribution is conceptual, but the current computational protocol prevents the central claims from being accepted as they stand. If the subunit-specific patterns survive the corrected ensemble treatment, the paper would be a valuable contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a genuinely new analysis of a well-established PSII rate matrix—no one has plotted the Shannon entropy and enthalpy components of the exciton population dynamics for these supercomplexes before. The population contraction time is a useful metric, and the contrast between D1-side and D2-side subunits is clean enough to be interesting even to someone who doesn't care about entropy. The paper is also honest about building on Yang et al. 2024 and Bennett et al. 2013, and the rate-matrix code is public.\n\nThe soft spots are real, though. The Methods explicitly say the entropy is calculated from the averaged rate matrix over 500 disorder realizations. That means they compute e^{⟨K⟩t}P(0) and then S(⟨p(t)⟩), not the proper ensemble average ⟨S(p_δ(t))⟩_δ or even S(⟨p(t)⟩) from ⟨e^{K(δ)t}⟩. Since rate constants depend nonlinearly on site-energy disorder, the averaged matrix is not representative of any complex, and Shannon entropy is concave, so Jensen gives a systematic upward bias. No error bars or distributions are shown for any contraction time. Because the subunit-role conclusions and the C2S2 vs C2S2M2 comparisons rest on these times, this is a load-bearing problem, not a stylistic one. The qualitative pattern—D1 side faster, D2 side more bidirectional, M-LHCII enabling inter-monomer transfer—matches prior work and experiments, so I suspect the story survives a correct averaging, but as written the central numbers are not supported.\n\nThe second soft spot is the thermodynamic interpretation. The authors define ΔS as the Shannon entropy of the exciton population vector and justify it via detailed balance. That's a legitimate coarse-grained information entropy, but it is not the physical entropy of the pigment-protein system, and for a system where coherence or non-Markovian phonon effects matter the classical rate picture is an approximation. The paper should say this explicitly rather than presenting ΔG = ΔH - TΔS as the free energy of the complex.\n\nThird, the 'design principle' language is teleological. The model shows patterns; it doesn't prove selection for entropy. That's a framing issue, and an easy fix.\n\nVerdict: this deserves peer review, not desk rejection. It's a solid computational study with a new angle and public code. But the authors should be asked to recompute with disorder-averaged dynamics and report distributions, to caveat the entropy interpretation, and to soften the design language. The reader's conditional verdict is right; I'd push slightly harder on the averaging issue, but the paper is worth engaging with.","headline":"New entropy decomposition of PSII dynamics is suggestive, but the disorder-averaging method undermines the quantitative claims; needs revision before the design-principle story can be trusted.","tokens_in":12319,"tokens_out":3452,"would_cite":true,"duration_ms":29828,"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":"Photosystem II reconciles efficient trapping and photoprotection by giving each protein subunit its own entropy–enthalpy driving profile, with entropy tunable through LHCII binding.","keywords":["photosystem II","excitation energy transfer","entropy","free energy landscape","LHCII antenna","nonphotochemical quenching","population contraction time","stochastic thermodynamics"],"falsifier":"A direct test would be time-resolved fluorescence or two-dimensional electronic spectroscopy on PSII that checks whether energy transfer from an LHCII chlorophyll shows the predicted two-phase behavior, with sub-picosecond entropic spreading followed by contraction on the tens-of-picoseconds scale, and whether removing M-LHCII complexes changes inter-monomer trapping rates in the predicted direction when one reaction center is closed. A clear mismatch in those timescales or in the direction of the M-LHCII effect would falsify the claim that entropy differences among subunits are the controlling design variable.","tokens_in":11278,"feed_emoji":"🌿","tokens_out":9095,"duration_ms":76149,"temperature":0.7,"pith_summary":"Photosystem II must do two opposing jobs: capture light with near-perfect efficiency and safely dissipate excess energy as heat. This paper argues that it accomplishes both by using entropy, not just energy, as a design variable. Simulating exciton relaxation on a structure-based rate matrix, the authors find that each protein subunit has a characteristic entropy–enthalpy balance: D1-side subunits drive directed, enthalpic transfer toward the reaction center, while D2-side and peripheral subunits spread excitations entropically, providing alternate pathways and inter-monomer routes. They further show that adding or removing LHCII antenna complexes tunes the entropic part of the free energy landscape, letting the complex adapt to light conditions. If correct, this reframes the flat PSII energy landscape as a feature rather than a puzzle.","feed_headline":"Entropy steers Photosystem II's light-harvesting design","feed_subtitle":"Each antenna subunit plays its own free-energy role, letting PSII trap light efficiently and shed excess heat.","key_machinery":"The central machinery is a structure-based exciton rate matrix $\\mathbf{K}$ built from the crystal structures of the C2S2 and C2S2M2 supercomplexes, with rates computed using quantum rate theory and organized into domains by coupling strength. Exciton populations evolve as $\\mathbf{P}(t) = e^{\\mathbf{K}t}\\mathbf{P}(0)$, and because the rates satisfy detailed balance with respect to a Boltzmann distribution, the authors interpret the Shannon entropy of the population distribution as thermodynamic entropy. The free energy decomposition is $\\Delta G(t) = \\Delta H(t) - T\\Delta S(t)$, and the population contraction time, defined as the time when entropy reaches its maximum, is the key observable used to classify subunits.","core_discovery":"The central claim is that the free energy landscape of the PSII supercomplex is organized so that entropy and enthalpy play different, complementary roles in different protein subunits, and this division of labor reconciles efficient trapping with photoprotection. The paper computes, for each initial excitation, the time-dependent entropy change $\\Delta S(t) = -k_B \\sum_i p_i(t) \\ln p_i(t)$ and enthalpy change $\\Delta H(t) = \\sum_i p_i(t) E_i - E(0)$, and finds two dynamical phases: an early entropy-driven phase in which population spreads among near-degenerate states, and a later enthalpy-driven contraction toward the reaction center. The time of maximum entropy (the population contraction time) varies by subunit: fast contraction with high same-monomer trapping for CP43, CP26, and S-LHCII on the D1 side; slower, bidirectional spreading for CP47, CP29, and M-LHCII on the D2 side; and long uphill excursions at Chls a 610 and 612 in LHCII, proposed nonphotochemical quenching sites. When one reaction center is closed, D1-side states in that monomer show disproportionately elongated contraction times, keeping excitations available to reach the other open RC. The authors conclude that the entropic component is not fixed but is dynamically regulated by LHCII binding, making entropy a tunable design principle.","pith_inferences":["The paper leaves implicit that the entropy-dominated spreading phase is a general strategy for robustness: any energy transfer network with many near-degenerate states could use entropy to maintain multiple pathways when one route is blocked.","The population contraction time could serve as a practical design metric for artificial systems, screening architectures for fast contraction under normal operation and slow contraction under stress.","Because the entropy term is derived from the probability distribution, the framework naturally connects to information-theoretic measures such as entropy production or mutual information between excitation and trap state.","A testable extension would be to mutate or reconstitute PSII with altered LHCII stoichiometry and measure whether the predicted shifts in population contraction times and same-monomer trapping probabilities appear in fluorescence kinetics."],"forward_implications":["Because each subunit has its own entropy–enthalpy balance, PSII does not need an energy funnel; the flat landscape works through subunit-specific driving forces.","Plants can dynamically retune the entropy component by adding or removing LHCII complexes, changing connectivity and inter-monomer transfer without changing pigment energies.","Closing one reaction center does not cripple trapping: D1-side excitations in the closed monomer slow down, leaving more time to reach the other open RC, and the slowest population contraction stays below the fluorescence lifetime.","The long-lived uphill states at Chls a 610 and 612 in LHCII provide a concrete molecular target for nonphotochemical quenching.","Artificial light-harvesting designs can use these principles: fast directed pathways for trapping, plus redundant entropy-rich pathways for backup under variable conditions."],"supporting_citations":[{"why":"Provides the kinetic transition network trajectories and inter-monomer transfer analysis that this work complements with ensemble free-energy terms.","marker":"[12]"},{"why":"Supplies the bidirectional energy-flow picture in the PSII supercomplex that motivates the D1/D2 asymmetry analysis.","marker":"[13]"},{"why":"Establishes the structure-based model of PSII energy transfer that the rate-matrix construction builds on.","marker":"[16]"},{"why":"Provides the C2S2 spinach supercomplex structure used for the smaller-complex rate matrix.","marker":"[19]"},{"why":"Provides the C2S2M2 pea supercomplex structure (PDB 5XNL) used for the main calculations.","marker":"[20]"},{"why":"Supplies the core complex Hamiltonian and trapping analysis underlying the D1-side faster trapping rates.","marker":"[27]"},{"why":"Gives the quantum rate theory used to compute intra- and inter-domain transfer rates for the matrix.","marker":"[32]"},{"why":"Justifies the thermodynamic interpretation of the rates via detailed balance and the free-energy decomposition.","marker":"[33]"},{"why":"Locates the charge-transfer state between ChlD1 and PheoD1, defining the trap position in the model.","marker":"[34]"},{"why":"Identifies Chls a 610 and 612 in LHCII as candidate nonphotochemical quenching sites, which the long uphill retention supports.","marker":"[35]"}],"fun_headline_variants":["Each PSII subunit plays its own entropy-enthalpy role","Entropy as a tunable knob in Photosystem II's antenna","How Photosystem II uses entropy to balance light and heat","Entropy and enthalpy split duties in Photosystem II","Entropy makes Photosystem II's light harvesting tunable"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the Shannon entropy of the exciton population distribution is the true thermodynamic entropy of the transfer process; this requires the rates to obey detailed balance with a Boltzmann distribution and the dynamics to be effectively Markovian, so any substantial coherence or memory effects in real PSII would break the entropy-based conclusions.","fun_headline_variants_meta":{"raw":{"variants":["Each PSII subunit plays its own entropy-enthalpy role","Entropy as a tunable knob in Photosystem II's antenna","How Photosystem II uses entropy to balance light and heat","Entropy and enthalpy split duties in Photosystem II","Entropy makes Photosystem II's light harvesting tunable"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000824,"raw_usage":{"total_tokens":3671,"prompt_tokens":1077,"completion_tokens":2594,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":693,"completion_tokens_details":{"reasoning_tokens":2511}},"tokens_in":693,"tokens_out":2594,"duration_ms":17187,"temperature":1.0,"reasoning_tokens":2511,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T14:05:48.303121+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct test would be time-resolved fluorescence or two-dimensional electronic spectroscopy on PSII that checks whether energy transfer from an LHCII chlorophyll shows the predicted two-phase behavior, with sub-picosecond entropic spreading followed by contraction on the tens-of-picoseconds scale, and whether removing M-LHCII complexes changes inter-monomer trapping rates in the predicted direction when one reaction center is closed. A clear mismatch in those timescales or in the direction of the M-LHCII effect would falsify the claim that entropy differences among subunits are the controlling design variable.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the kinetic transition network trajectories and inter-monomer transfer analysis that this work complements with ensemble free-energy terms."},{"cited_title":"Leonardo et al., Bidirectional energy flow in the photo- system ii supercomplex, The Journal of Physical Chem- istry B 128, 7941 (2024)","cited_arxiv_id":null,"evidence_quote":"Supplies the bidirectional energy-flow picture in the PSII supercomplex that motivates the D1/D2 asymmetry analysis."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the structure-based model of PSII energy transfer that the rate-matrix construction builds on."},{"cited_title":"Wei et al","cited_arxiv_id":null,"evidence_quote":"Provides the C2S2 spinach supercomplex structure used for the smaller-complex rate matrix."},{"cited_title":"Su et al., Structure and assembly mechanism of plant c2 s2 m2-type psii-lhcii supercomplex, Science 357, 815 (2017)","cited_arxiv_id":null,"evidence_quote":"Provides the C2S2M2 pea supercomplex structure (PDB 5XNL) used for the main calculations."},{"cited_title":"Raszewski and T","cited_arxiv_id":null,"evidence_quote":"Supplies the core complex Hamiltonian and trapping analysis underlying the D1-side faster trapping rates."},{"cited_title":"Yang and G","cited_arxiv_id":null,"evidence_quote":"Gives the quantum rate theory used to compute intra- and inter-domain transfer rates for the matrix."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Justifies the thermodynamic interpretation of the rates via detailed balance and the free-energy decomposition."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Locates the charge-transfer state between ChlD1 and PheoD1, defining the trap position in the model."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Identifies Chls a 610 and 612 in LHCII as candidate nonphotochemical quenching sites, which the long uphill retention supports."}],"review_version":1}