{"id":"228e0147-2f21-43df-a2f5-6bbf7ecbb794","arxiv_id":"2601.15477","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"For strongly deformed membranes, protein inclusions feel non-monotonic forces and interact with a sub-power-law distance dependence, with a flow-speed threshold for deformation.","lead":"This paper uses computer simulations and a simple formula to predict how strongly bent cell membranes push and pull on proteins embedded in them. It finds that forces become non-monotonic at large deformations, and that flowing membranes may start to deform only above a certain speed.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Flow threshold v*=κ/(Lη) is set by the simulation-box size; as L→∞, v*→0, so the claimed onset of flow-induced deformation is a finite-size artifact rather than an intrinsic physical scale.","rationale":"The reader's weakest assumption correctly identifies the flow threshold as L-dependent, and this is the single most load-bearing concern because it directly undermines one of the four central claims in the abstract (the characteristic velocity/onset of flow-induced deformation). The concern is not merely a matter of fitting: the threshold v*=κ/(Lη) vanishes in the infinite-membrane limit, and the paper's own derivation is circular—the boundary conditions (34)-(35) impose flat membrane at distance L, so any deformation with wavelength λ>L is artificially suppressed. A domain-size study would settle whether the onset is intrinsic or an artifact. The other results (non-monotonic force, pair interaction, charge analogy) are less affected because they are computed under fixed boundary conditions without invoking a threshold; however, they too may carry finite-size effects, but the reader's request for convergence data is secondary. Since the paper's overall contribution can be salvaged by reinterpreting the flow result as a finite-size phenomenon and the remaining claims are credible, the appropriate verdict remains CONDITIONAL, matching the reader's. No evidence of internal inconsistency or fraud was found; the zero-mean-curvature analytical solution is exact in the H=0 sector and the code (IRENE) is described in a separate preprint, providing independent support for the numerical framework.","tokens_in":12429,"tokens_out":6869,"duration_ms":76508,"concrete_test":"Repeat the flow simulation for a single PI in a square domain at a fixed inflow velocity v0 = 0.5 κ/(100r0 η), which is below the v* reported for L=100r0. Run the same setup for L = 100r0, 200r0, 400r0, and 800r0, and measure the maximum steady-state membrane displacement. If a nonzero deformation appears for larger L (where v0 > κ/(Lη)), the 'no-deformation' regime at L=100r0 is a finite-size artifact. A complementary check: compute v* for three domain sizes and verify whether v* ∝ 1/L; if the proportionality holds exactly, the threshold is not an intrinsic physical scale but a box-size-dependent cutoff.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing concern is the claimed flow-induced deformation threshold. In §III.B, the outer boundary is imposed flat via Eqs. (34)-(35) at distance L, and the threshold is defined in Eq. (37) as v*=κ/(Lη). This definition makes the onset explicitly dependent on the numerical domain: as L→∞, v*→0, so for any nonzero flow velocity an infinite membrane would exhibit a deformation with wavelength λ=κ/(ηv), and the 'no-deformation' regime disappears. The paper's own explanation (end of §III.B) is circular: because deformations are assumed not to propagate beyond L (the boundary conditions (34)-(35) fix z=0 and zero normal slope at ∂Ω), no deformation appears when λ>L. Hence v* is a boundary-condition artifact rather than a material property. Consequently, the abstract's claim (iv)—a characteristic velocity separating bending- and viscous-dominated regimes with an onset of flow-induced deformation—is unsupported in the present form. The other central claims (non-monotonic force, sub-power-law pair interaction, charge analogy) are numerical results computed for fixed boundary conditions and do not rely on this threshold, so they remain credible; the flow section needs major revision or re-scoping as a finite-size effect.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies membrane-protein interactions in the large-deformation (LD) regime using finite-element simulations performed with the authors' IRENE software. It claims four main results: (i) the membrane force on a single protein inclusion is non-monotonic in the vertical displacement; (ii) for two inclusions the membrane-mediated potential decays slower than a power law with distance; (iii) conical inclusions with equal and opposite orientations repel and attract, respectively, confirming the 'charge analogy' beyond the small-deformation limit; and (iv) in the presence of membrane flow, a characteristic velocity v* = κ/(Lη) separates a no-deformation regime from a flow-deformation regime. The paper also presents an approximate zero-mean-curvature analytic solution and compares it with numerical solutions.","tokens_in":12715,"tokens_out":5004,"duration_ms":53663,"significance":"If the results hold, the paper would extend membrane-mediated protein interactions beyond the well-studied small-deformation perturbative regime, and it provides concrete quantitative predictions for force-displacement curves, pair potentials, and flow-induced shape changes. Strengths of the paper include the use of a full nonlinear shape equation, the explicit zero-mean-curvature analytic solution (exact when H = 0), and the fact that the numerical results are in principle reproducible through the companion IRENE software [28]. The GUV/bacteriorhodopsin example makes the predictions biologically concrete. However, the numerical claims are not fully supported by in-paper convergence/error analysis, and the flow-velocity threshold is, as discussed below, tied to the computational domain size.","major_comments":[{"comment":"The characteristic velocity v* is defined as v* = κ/(Lη), where L is the simulation-box side. The outer boundary conditions (34)-(35) impose z = 0 and zero normal slope at distance L. As the manuscript itself explains at the end of §III.B, a flow-induced deformation with wavelength λ = κ/(ηv) can appear only when λ < L; otherwise it is suppressed by the boundary. Consequently, v* depends explicitly on the computational domain and v* → 0 as L → ∞. For a truly infinite membrane, any nonzero v would produce a deformation with finite wavelength, so the claimed 'no-deformation' regime and the onset threshold are finite-size artifacts rather than an intrinsic physical scale. This undermines central claim (iv) of the abstract and the corresponding discussion in §IV. The section should be re-scoped as a finite-size effect or supplemented by an L-dependence study and a physical argument for a fin","section":"§III.B, Eq. (37), Eqs. (34)-(35)"},{"comment":"The non-monotonic force is one of the paper's central predictions, but Figure 5 shows numerical curves without error bars or mesh-convergence metrics. The text states that a convergence analysis for the radially symmetric case is reported in the companion paper [28], but the present manuscript does not provide it. For a quantitative claim about the magnitude, sign, and zero crossings of the force, representative convergence data or an error estimate should be included in this paper, or the claim should be explicitly labeled as preliminary.","section":"§III.A.3, Fig. 5"},{"comment":"The claim that the two-protein interaction potential exhibits 'sub-power-law decay' is based on visual curvature in a log-log plot. No fit to a power law is shown, no slope is computed, and no alternative decay form (e.g., logarithm, stretched exponential) is tested. Since this is a central quantitative result, the manuscript should define 'sub-power-law' quantitatively and demonstrate that the data are inconsistent with a power-law over the presented range, accounting for numerical uncertainty.","section":"§III.A.4, Fig. 8A"},{"comment":"The derivation of Eq. (10) from Eq. (1) is not shown; the text simply states that substituting Eq. (A1) into Eq. (1) gives Eq. (10). A reader cannot verify the algebra, and the regime of validity of the approximate equation is therefore unclear. In addition, Eq. (11) defines H = ∂ψ/∂r + ψ/r, while Appendix A gives H = (1/2)(∂ψ/∂r + ψ/r). The factor 1/2 is immaterial for the zero-mean-curvature solution ψ = C1/r, but the inconsistency is confusing and should be clarified.","section":"§III.A.2, Eq. (10), Appendix A, Eq. (A1)"}],"minor_comments":[{"comment":"The caption says '0 ≤ x2 ≤ h' but the domain side length is presumably L, not h. This appears to be a typo.","section":"Fig. 6 caption"},{"comment":"The parameter t is introduced in Eq. (33) and used in Fig. 10, but it is not defined in Table I or in the text. Please define t (presumably tanα at the protein boundary).","section":"Eq. (33) and Fig. 10"},{"comment":"The notation ∂Ω in Eq. (36) as a union of three boundaries is confusing because the same symbol ∂Ω is also used for the outer boundary in Eqs. (34)-(35). Distinct symbols for the inner and outer boundaries would improve readability.","section":"Eq. (36), Fig. 9"},{"comment":"Eq. (16) is stated as the R≫r0 limit of Eq. (15), but the derivation is not shown; expanding Eq. (15) gives the logarithmic terms, but the cross-check of signs and the domain of validity of the expansion should be stated.","section":"Eq. (15) and Eq. (16)"},{"comment":"The viscosity unit 'Pa m sec' should be 'Pa·m·s' to indicate a 2D membrane viscosity; this is not a substantive issue but would avoid confusion.","section":"Table I"}],"recommendation":"major_revision","confidential_remarks":"The flow-threshold issue is the most serious problem and appears to be a finite-size artifact. However, it is localizable and could be fixed by re-scoping or by an explicit L-dependence study; therefore major revision, not rejection, is appropriate. The other central claims are plausible but would be more convincing with in-paper convergence data, especially since all numerical solutions come from the authors' own IRENE software. In addition to the major comments, I would encourage the editor to request a short statement from the authors about the validation of IRENE against independent solutions or benchmarks."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The real news here is the finite-element work: first full solutions for finite-size protein inclusions in the large-deformation regime, with a non-monotonic force–displacement curve and a two-inclusion interaction potential that decays slower than a power law. Those results look plausible and are a genuine step beyond the small-deformation literature. The analytic zero-mean-curvature solution (Eq. 14) is exact when H=0 and gives a useful closed-form approximation for large contact angles. The force profiles in Fig. 5 are the strongest part, and the charge analogy surviving beyond small deformations is a nice confirmation. I would not be surprised if the pair-interaction curves become a reference for future work.\n\nThe soft spots are mostly in the flow section, and the stress-test note is right. The threshold v*=κ/(Lη) is defined with L as the simulation box size, and the boundary conditions fix the membrane flat at that distance. So v* is not a material property; as L→∞ it goes to zero, and in an infinite membrane any nonzero flow has a deformation wavelength λ=κ/(ηv). The paper's own explanation—deformations with wavelength larger than L are suppressed by the boundary—is circular. That claim in the abstract, about a characteristic velocity separating bending- and viscous-dominated regimes, needs to be re-scoped as a finite-size threshold or dropped. The rest of the paper does not depend on this threshold, so the force and interaction results stand.\n\nOther issues are minor by comparison. The derivation of Eq. (10) is not shown, and the numerical curves in Figs. 5, 8, and 11 have no error bars or convergence metrics; the authors defer to the companion IRENE paper, which is acceptable but should be made explicit. The “sub-power-law” claim lacks a quantitative baseline—what power law is it being compared to? A fitted exponent or a comparison to the known small-deformation scaling would settle it. The citation pattern looks fine, and using their own software is not a problem if it is public and validated.\n\nBottom line: this is a solid contribution on the mechanics side, with one overstated claim. A serious referee should engage with it and ask for the flow section to be revised or relabeled, and for numerical convergence data. It deserves peer review, not desk rejection.","headline":"Useful large-deformation FE results and a clean analytic minimal-surface solution, but the flow threshold v*=κ/(Lη) is a finite-box artifact and the paper overclaims it as an intrinsic scale.","tokens_in":13195,"tokens_out":2412,"would_cite":true,"duration_ms":28044,"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":"This paper establishes that beyond small membrane slopes, a protein inclusion feels a non-monotonic restoring force, pairwise interactions decay slower than a power law and keep a charge-like sign rule, and a flow speed v* = κ/(Lη) marks th","keywords":["lipid membranes","protein inclusions","large deformations","Helfrich energy","finite element method","membrane-mediated interactions","conical inclusions","membrane flow"],"falsifier":"Solve the flow equations for the same protein and material parameters with the outer box size L doubled and tripled at a fixed inflow velocity; if the onset of deformation does not move as v* ∝ 1/L—meaning a deformation appears at a velocity that was previously sub-threshold—the flow-onset claim is contradicted.","tokens_in":12274,"feed_emoji":"🔬","tokens_out":9274,"duration_ms":101838,"temperature":0.7,"pith_summary":"Most theoretical work on proteins bending cell membranes has lived in the small-deformation regime, where the membrane slope is treated as tiny. This paper shows what happens when that assumption breaks: in the large-deformation regime, solved with a nonlinear finite-element method, the membrane's restoring force on a protein inclusion is non-monotonic in the inclusion's vertical displacement, the membrane-mediated interaction between two inclusions decays slower than a power law with distance, and conical inclusions with the same orientation repel while opposite orientations attract, extending the 'charge' picture. It also shows that a background membrane flow only begins to visibly deform the membrane once the flow speed exceeds v* = κ/(Lη), with a deformation wavelength of about κ/(ηv). The payoff is a set of quantitative predictions for strongly deformed membrane-protein systems—relevant to protein sorting, clustering, and membrane trafficking—plus an approximate closed-form membrane profile for quick estimates.","feed_headline":"Protein forces go non-monotonic at large membrane bends","feed_subtitle":"Beyond small deformations, curvature charges still attract and repel, and fast flow bends membranes.","key_machinery":"The engine is the full, nonlinear Helfrich shape equation—the Euler–Lagrange equation of the standard bending-plus-tension energy—solved by finite elements rather than by expanding about a flat membrane. The main analytic device is a change of slope variable ω = ψ/√(1−ψ²); assuming zero mean curvature gives ψ = C/r and a logarithmic membrane profile z(r) = −r0 sinα ln[(r−√(r²−r0² sin²α))/(R−√(R²−r0² sin²α))], a closed-form map from contact angle to spontaneous displacement. The force calculation comes from the boundary terms of the varied Helfrich functional, giving normal and tangential line forces f⊥ = −2κ ∂H/∂r and f∥ = −n_i(2κH² + σ)e_i. The two-protein results come from numerically eval","core_discovery":"The central discovery is that the qualitatively new features of protein–membrane interaction appear only when the membrane is strongly deformed, and they can be captured numerically. Linearized small-deformation theory is adequate only for contact-angle slopes up to about tan α ≈ 0.4; beyond that the full shape equation is needed. In this large-deformation regime the vertical membrane force on a single protein is non-monotonic in the protein's displacement; the two-protein interaction energy decays sub-power-law with separation, and its sign follows the 'charge' rule of the conical contact angles. In the presence of flow, the paper finds a crossover speed v* = κ/(Lη) below which flow has no","pith_inferences":["Reading the boundary conditions literally, the flow threshold v* = κ/(Lη) scales inversely with the simulation-box size L, so for an unbounded membrane the threshold would tend to zero; the robust physical scale is the deformation wavelength λ = κ/(ηv), and the apparent 'onset' is likely a finite-domain crossover rather than an intrinsic biological speed.","Since small-deformation studies already show non-pairwise forces among multiple inclusions, the sub-power-law two-body potential found here suggests that clusters of three or more proteins in the large-deformation regime may develop many-body ordering not captured by pairwise sums—an untested but natural extension.","The non-monotonic force curve offers a concrete experimental target: a micropipette or optical-tweezer measurement of force versus displacement on a vesicle with a protein inclusion should reveal a force peak that shifts with the patch size, directly testing the large-deformation prediction.","The closed-form logarithmic profile could serve as a cheap building block for coarse-grained or multi-protein screening models, letting researchers explore LD effects without a finite-element solve for every configuration."],"forward_implications":["At large vertical displacements the membrane's restoring force on a protein stops growing and falls, so strongly invaginated inclusions are pulled back less forcefully; this changes estimates of the forces needed for endocytic uptake or membrane tube formation.","Membrane-mediated interactions between two inclusions decay more slowly than a power law at the distances studied, so at separations of a few protein radii the interaction is stronger than small-deformation theory would predict, affecting clustering and pattern formation.","The charge analogy survives large deformations: two conical inclusions with the same orientation repel, opposite orientations attract, so a mixture of differently oriented curvature-generating proteins will tend to segregate by orientation.","A flow speed above v* = κ/(Lη) deforms the membrane with wavelength ~κ/(ηv); for typical protein diffusion velocities this wavelength can be microns, comparable to inter-protein spacings, so flow-induced deformation can affect protein mobility and organization in dense arrays.","The approximate zero-mean-curvature profile provides a ready formula for the spontaneous displacement of a protein with a given contact angle, letting experimentalists estimate displacements without solving the full equations."],"fun_headline_variants":["Membrane forces on proteins turn non-monotonic at large bends","Curvature charges attract and repel even under extreme membrane bends","Sub-power-law decay marks protein interactions in bent membranes","Flow deforms membranes only above a critical velocity","Extreme membrane bending exposes new protein interaction laws"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise in the flow section is that the membrane is flat and undisturbed at the outer boundary of the simulation box, so the threshold speed v* = κ/(Lη) shrinks as the box grows and would vanish for an unbounded membrane—if that boundary condition is relaxed, the flow-onset claim may not hold.","fun_headline_variants_meta":{"raw":{"variants":["Membrane forces on proteins turn non-monotonic at large bends","Curvature charges attract and repel even under extreme membrane bends","Sub-power-law decay marks protein interactions in bent membranes","Flow deforms membranes only above a critical velocity","Extreme membrane bending exposes new protein interaction laws"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000453,"raw_usage":{"total_tokens":2123,"prompt_tokens":759,"completion_tokens":1364,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":503,"completion_tokens_details":{"reasoning_tokens":1285}},"tokens_in":503,"tokens_out":1364,"duration_ms":23699,"temperature":1.0,"reasoning_tokens":1285,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T08:50:57.400420+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Solve the flow equations for the same protein and material parameters with the outer box size L doubled and tripled at a fixed inflow velocity; if the onset of deformation does not move as v* ∝ 1/L—meaning a deformation appears at a velocity that was previously sub-threshold—the flow-onset claim is contradicted.","supporting_citations":[],"review_version":1}