{"id":"0143c396-d3c0-4535-ae6d-c18c0b64521f","arxiv_id":"2508.18570","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Comparing force fields on EK polyampholytes against SAXS data finds that ff19SB with OPC water matches experiments best, showing the protein force field matters more than previously thought.","lead":"The authors ran long atomistic simulations of three EK polyampholyte sequences with three protein/water force-field combinations and compared the computed small-angle X-ray scattering profiles against experimental data. They conclude that AMBER ff19SB with OPC water best reproduces both ordered and disordered ensembles, and they release a new open-source tool for computing scattering from any MD engine.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Single-trajectory sampling is the load-bearing risk: the narrow (E4K4)4 Rg peak under FF19O may be a metastable helical basin, so the ordered/disordered generalizability claim is not yet settled.","rationale":"The paper is technically careful: SWAXS-AMDE is validated against lysozyme (SI S.3), the code is open source, the explicit-solvent scattering treatment is a reasonable one-to-one comparison tool, and the force-field decomposition (TFF99/OFF99/FF19O) is a sensible design. I do not question the experimental SAXS data or the software itself. The concern is the inferential leap from three single trajectories to a generalized model. The reader's weakest-assumption diagnosis matches mine: equilibrium is asserted rather than demonstrated for the very system where the headline behavior appears. A narrow Rg distribution is exactly what a kinetically trapped ordered state looks like, and starting from one GB/SA-derived conformation does not protect against this. The first-half/second-half Ramachandran comparison in SI S.6 is necessary but not sufficient. A decisive test is to restart (E4K4)4 with FF19O from structurally diverse initial conformations and/or use replica exchange; this directly tests whether the 0.27 A width is a free-energy basin or a kinetic bottle. Until then the appropriate verdict remains conditional: the data support ff19SB-OPC as promising, not as an established generalizable model. No change to the reader's conditional verdict is needed.","tokens_in":21864,"tokens_out":4562,"duration_ms":46288,"concrete_test":"Run temperature replica-exchange MD, or at minimum three independent 2-4 microsecond MD starts from extended, collapsed, and helical initial conformations, for FF19O (E4K4)4 using the same 293.15 K, NPT, SWAXS-AMDE protocol. Recompute the Rg distribution and the calculated SAXS profile. If the narrow peak at 14.96 +/- 0.27 A and chi about 1.40 are reproduced in every start or replica, metastability is ruled out; if the distribution broadens or shifts, the ordered-state prediction and the generalizability claim are not established.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing premise is that a single 8 microsecond trajectory per system, started from one GB/SA-derived conformation (SI S.1), has equilibrated the conformational ensemble. The convergence check in SI S.6 (Ramachandran similarity between first and second halves) only shows that each trajectory stopped changing character; it cannot distinguish equilibrium from a trapped metastable basin. This matters most for FF19O (E4K4)4, where Tab. S1 reports mean Rg = 14.96 A with std = 0.27 A and Fig. 4(I) shows a highly populated alpha-R basin. If that trajectory is kinetically trapped in a helix, the 'ordered polyampholyte' prediction that underpins the generalizability claim is an artifact. The central FF19O-versus-OFF99 ranking in Fig. 2 and Tab. S2 would also be compromised, because the comparison would then be against the wrong FF19O ensemble. Since the claim is explicitly about a generalized model, one trajectory per sequence is the weakest support. The paper itself flags in Sec. 3.3 that secondary-structure predictions have not yet been tested by NMR/CD/FTIR, leaving the proposed mechanism unverified.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper tests three AMBER force-field/water combinations (ff99SB-TIP3P, ff99SB-OPC, ff19SB-OPC) against SAXS data for three EK polyampholyte sequences: (EK)16, (E2K2)8, and (E4K4)4. Each system is simulated for 8 µs, and scattering is computed with a new explicit-solvent model (SWAXS-AMDE) that accounts for hydration-layer density and thermal fluctuations. The authors report that ff19SB-OPC gives the lowest χ values for two of the three sequences, yields a wide Rg distribution for (E2K2)8 and a narrow one for (E4K4)4, and therefore claim that ff19SB-OPC is a generalized model for both disordered and ordered polyampholytes, with the protein force field playing a central role.","tokens_in":22105,"tokens_out":5349,"duration_ms":47455,"significance":"If the conclusions hold, the paper makes a useful contribution by showing that backbone dihedral corrections in the protein force field, not only water-model dispersion, control IDP ensemble quality, and by providing an open-source, engine-agnostic scattering tool. The design is strong in several respects: 8 µs trajectories are longer than typical IDP validation runs; the scattering comparisons include block-averaged errors; SWAXS-AMDE is validated against lysozyme and released on GitHub; and no conformational parameter is fitted to the SAXS data. The main limitation is statistical: one trajectory per system cannot by itself establish equilibrium, and the key ordered-sequence result (FF19O for (E4K4)4) may reflect a metastable helical basin.","major_comments":[{"comment":"The convergence check is not sufficient to support the equilibrium assumption. SI S.6 reports that the Ramachandran plots from the first 4 µs and the second 4 µs 'characteristically match,' but comparing two halves of a single trajectory only demonstrates stationarity within that trajectory, not convergence to the equilibrium ensemble. The strongest evidence of a problem is FF19O for (E4K4)4: Tab. S1 reports mean Rg = 14.96 Å with a standard deviation of 0.27 Å, and Fig. 4(I) shows a dominant αR basin. A single 8 µs trajectory started from one GB/SA-derived conformation can be kinetically trapped in a helical basin, and if that is the case, the ordered/disordered distinction that underpins the generalizability claim would be an artifact. I recommend adding independent trajectories from different starting conformations, replica-exchange or other enhanced-sampling runs, or at minimum a direct analysis of helix-coil transition rates and basin escape times; otherwise the central claim should be explicitly conditioned on this assumption.","section":"Sec. 2.3, SI S.6, Tab. S1, Fig. 4"},{"comment":"The claim that FF19O is the best performing model is not uniform across the three sequences. Tab. S2 shows FF19O achieves the lowest χ for (EK)16 (2.00 vs 3.77 for OFF99) and (E4K4)4 (1.40 vs 5.96), but for (E2K2)8, OFF99 gives χ = 1.92 while FF19O gives 2.06. Since (E2K2)8 is the sequence that most directly probes disordered, IDP-like behavior, this reversal needs to be acknowledged and discussed; the conclusion in Sec. 3.1 and Sec. 4 that 'FF19O performed the best' is therefore only true for two of the three systems, and the claimed superiority of ff19SB over ff99SB for this disordered case is not supported by the reported numbers.","section":"Tab. S2, Sec. 3.1"},{"comment":"The residual disagreement at q > 0.3 Å−1 is treated as a future NMR/CD/FTIR question, but it deserves a quantitative discussion in the present work. The high-q region reports local backbone and hydration structure, so the systematic mismatch shown in Fig. 2 for all three FF19O profiles is directly relevant to the mechanism proposed in Sec. 3.3 (that the dihedral-angle populations, in particular the αR dominance for (E4K4)4, explain the favorable comparison). Without an assessment of whether another secondary-structure composition would reduce the high-q mismatch, the mechanistic link between the Ramachandran populations and the experimental scattering remains a hypothesis rather than a demonstrated cause.","section":"Sec. 3.3, Fig. 2"}],"minor_comments":[{"comment":"The phrase 'without any free-parameters' is overstated. The χ2 minimization in Eq. (3) includes a scale factor f, and SI S.7 adds a constant offset c; although these parameters do not alter conformational sampling, they are fitted to the experimental scattering, so the statement should be qualified as 'without fitted conformational parameters.'","section":"Abstract, Sec. 2.4, Conclusions"},{"comment":"Equation (2) is typeset ambiguously in the main text; the ensemble-average brackets and the variance terms are difficult to parse. Please reformat the equation to match the clearer expression in SI S.2.","section":"Eq. (2)"},{"comment":"The software name appears inconsistently as 'SW AXS-AMDE' and 'SWAXS-AMDE'; please choose one spelling and use it consistently.","section":"Throughout"},{"comment":"The caption contains duplicated panel labels ('B) (E2K2)8 C) (E4K4)4' appears twice and a stray 'B)' appears in the middle), which makes the figure difficult to follow.","section":"Fig. 2 caption"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope and the scattering tool is a useful contribution. The main issue for the editor is statistical: the generalizability claim rests on one trajectory per sequence, and the narrow FF19O (E4K4)4 ensemble may be a trapped helical state. If the authors add replicate trajectories or enhanced-sampling evidence, the paper could be suitable; otherwise the claims should be substantially softened."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper does something worth doing: it isolates the protein and water model contributions in IDP simulations and validates ff19SB-OPC against SAXS for three EK polyampholytes. The genuinely new bit is the validation, not the formalism. The scattering equation is from Park et al. and Chen and Hub, and SWAXS-AMDE is a helpful multi-engine reimplementation rather than a new theory. That said, the code is open source, validated against lysozyme, and the authors are transparent about the residual high-q discrepancies. Those are real strengths.\n\nThe simulation setup is careful: 8 µs per system, block averaging, explicit water scattering with hydration-shell detail, and honest discussion of the free-parameter issue (the f and c in the χ comparison are standard, but calling the analysis “without free-parameters” overstates it slightly). The conclusion that ff19SB-OPC is a promising generalized model is plausible and supported by the Rg distributions and Ramachandran analysis.\n\nThe soft spots are real but mostly the usual ones for this type of study. The generalizability claim rests on three sequences and one trajectory each. The narrow Rg distribution for (E4K4)4 under FF19O (std 0.27 Å) really could be a kinetically trapped helical basin rather than the equilibrium ensemble. The convergence check in the SI only shows that the first and second halves of each trajectory look similar, which does not distinguish equilibrium from metastability. Also, FF19O is not uniformly best: for (E2K2)8 it is slightly worse than OFF99 (χ 2.06 vs 1.92), a detail the paper reports honestly but that weakens the broad-stroke claim. And the paper itself flags that secondary-structure predictions have not been checked against NMR/CD/FTIR.\n\nNone of this sinks the paper. The work is careful, the data are real, and the open-source tool will be useful to the IDP simulation community. But the title-level claim that the force field plays a crucial role is supported, while the “generalized model” claim needs more evidence. For a serious referee, I would ask for replicate simulations from different starting conformations and, ideally, a few more sequences or an independent observable. The paper deserves peer review; it is not a desk reject.\n\nOverall: a solid, useful contribution that should be published after the sampling concern is addressed or at least explicitly bounded in scope.","headline":"Careful force-field comparison with a useful open-source scattering tool, but the generalizability claim is stretched by single-trajectory sampling and three sequences.","tokens_in":22665,"tokens_out":1868,"would_cite":true,"duration_ms":19989,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["87.15.ap"],"model":"deepseek-v4-flash","headline":"AMBER ff19SB with OPC water predicts both ordered and disordered EK polyampholyte ensembles in agreement with SAXS experiments, making the protein force field, not just the water model, central to accurate ensembles.","keywords":["intrinsically disordered proteins","polyampholytes","force field validation","small-angle X-ray scattering","molecular dynamics","ff19SB","OPC water model","radius of gyration"],"falsifier":"Run FF19O simulations of (E4K4)4 from several very different starting structures, such as an extended chain and a compact coil, and compare the resulting radius-of-gyration distributions and SAXS curves; if the narrow distribution (mean about 14.96 Å, spread about 0.27 Å) widens substantially or no longer matches the experimental scattering, the equilibrium assumption fails.","tokens_in":21654,"feed_emoji":"🧬","tokens_out":10709,"duration_ms":95140,"temperature":0.7,"pith_summary":"The paper asks which ingredient—the protein force field or the water model—determines whether molecular dynamics simulations reproduce the solution ensembles of intrinsically disordered proteins. Using experimental small-angle X-ray scattering data for three sequence-defined EK polyampholyte peptides, the authors compare three force-field/water combinations and isolate the two contributions. They find that replacing TIP3P water with OPC water improves agreement, but replacing the protein force field as well gives the best match. Their central claim is that the AMBER ff19SB-OPC water combination is a generalized model: it predicts the disordered ensembles of two sequences and the stable, preferred conformation of a third in line with experiments. This matters because force fields tuned for folded proteins have historically produced overly compact IDP conformations, and a single transferable model would let simulations be trusted for both folded and disordered proteins.","feed_headline":"Force field choice drives accuracy of disordered protein ensembles","feed_subtitle":"SAXS comparisons show the protein model, not only the water model, decides whether simulations match ordered or disordered states.","key_machinery":"The load-bearing design is a three-way controlled comparison: TFF99 (ff99SB with TIP3P water), OFF99 (ff99SB with OPC water), and FF19O (ff19SB with OPC water). That ordering isolates the water-model change first and then the protein-force-field change, allowing the two effects to be separated. The scattering comparison is carried out by a new model, SWAXS-AMDE, which computes background-subtracted SAXS intensities from explicit-water trajectories in atomic detail, including hydration-layer density changes and thermal fluctuations of the solute, so the computed curves can be compared to experiments without free parameters. The conformational analysis uses radius-of-gyration distributions and Ramachandran plots to link secondary-structure populations—especially the balance of alpha-helix and beta-sheet backbone angles set by dihedral-angle correction maps (CMAPs)—to whether an ensemble is ordered or disordered.","core_discovery":"The central discovery is that the protein force field contributes as much as the water model to obtaining experimentally faithful ensembles of disordered proteins. By keeping the protein force field fixed and switching the water model, then switching only the protein force field, the authors isolate each effect. The ff19SB-OPC combination gives the lowest chi values against the SAXS data, reproduces the Guinier-derived radius of gyration for all three sequences, and is the only combination that predicts a narrow size distribution for (E4K4)4 while still predicting broad, disordered distributions for (EK)16 and (E2K2)8. The authors conclude that reweighting backbone dihedral angles with CMAPs does more than refine secondary structure; it can move the whole macromolecular ensemble into the correct sequence-dependent ordered or disordered regime.","pith_inferences":["A testable extension is to launch FF19O simulations of (E4K4)4 from several independent starting structures; if the radius-of-gyration distribution does not reproduce the narrow spread reported here, the ordered state would be a sampling artifact rather than an equilibrium prediction.","The same three-way comparison could be repeated with other recently tuned backbone parameters and other polyampholyte sequences to ask whether the real mark of generalizability is sequence sensitivity—distinct secondary-structure populations for different charge blockings—rather than matching the mean radius of gyration.","The paper's conclusion that protein and water effects cannot be separated by length scale implies that next-generation force-field parameterization should be validated against scattering profiles over a wide range of scattering vectors, not just scalar size measures.","If the ordered (E4K4)4 state is genuine, the block-length trend in the paper's Fig. 3D suggests longer EK blocks will show progressively narrower, sequence-specific size distributions; synthesizing longer block polyampholytes and measuring SAXS or FRET would test this."],"forward_implications":["If FF19O is as general as claimed, molecular dynamics simulations of disordered proteins can be used without post-hoc reweighting or free fitting parameters to decide whether a sequence forms an ordered or a disordered ensemble.","Force-field development should treat the protein backbone parameters and the water model as coupled: replacing water alone is not enough, because the same OPC water gives poorer agreement when paired with ff99SB than with ff19SB.","The computed scattering profiles and secondary-structure populations for the three EK sequences become testable predictions for independent NMR, CD, and FTIR measurements, which the authors state are planned.","Detailed, parameter-free scattering comparison should become the standard for force-field validation; continuum models with adjustable parameters can mask whether a force field is actually producing the correct ensemble.","Residual disagreement at scattering vectors above about 0.3 per angstrom indicates that the hydration layer, secondary-structure proportions, or both still need refinement even in the best model."],"supporting_citations":[{"why":"Supplies the sequence-defined EK polyampholyte synthesis, the SAXS data, and the Guinier analysis that the computed scattering profiles are tested against.","marker":"[24]"},{"why":"Proposed OPC water as a general-purpose water model for IDPs and introduced the OFF99 combination; the present work isolates and extends that claim.","marker":"[19]"},{"why":"Defines the ff19SB protein force field with solution-trained backbone parameters and CMAP dihedral corrections, the key protein-force-field improvement tested.","marker":"[47]"},{"why":"Supplies the ff99SB baseline force field used in the TFF99 and OFF99 arms of the comparison.","marker":"[45]"},{"why":"Introduced the OPC water model, whose increased water-water dispersion is the water-side change being tested.","marker":"[17]"},{"why":"Derives the explicit-water scattering equation that is the theoretical core of the new scattering model.","marker":"[33]"},{"why":"Establishes the icosphere-envelope method for thermally fluctuating solutes, the solvent density correction, and the frame-count guidance adopted here.","marker":"[35]"},{"why":"Provides the fluctuating-solute scattering methodology that the new model generalizes for use with any MD engine.","marker":"[34]"},{"why":"Shows that free parameters in continuum scattering models can cancel force-field errors, motivating the parameter-free comparison used here.","marker":"[54]"}],"fun_headline_variants":["Protein force field, not just water, decides IDP ensemble accuracy","FF19SB-OPC wins: protein force field key for disordered proteins","Protein model choice shapes disordered protein ensembles","Force field matters: IDP ensembles need right protein parameters","Disordered protein accuracy hinges on protein force field"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim rests on the assumption that a single 8-microsecond simulation starting from one initial conformation reaches the true equilibrium ensemble, even for (E4K4)4, whose simulated size barely fluctuates; if that trajectory is actually trapped in a single shape, the ordered/disordered distinction and the generalizability claim would be artifacts.","fun_headline_variants_meta":{"raw":{"variants":["Protein force field, not just water, decides IDP ensemble accuracy","FF19SB-OPC wins: protein force field key for disordered proteins","Protein model choice shapes disordered protein ensembles","Force field matters: IDP ensembles need right protein parameters","Disordered protein accuracy hinges on protein force field"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00061,"raw_usage":{"total_tokens":2882,"prompt_tokens":1030,"completion_tokens":1852,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":646,"completion_tokens_details":{"reasoning_tokens":1771}},"tokens_in":646,"tokens_out":1852,"duration_ms":13864,"temperature":1.0,"reasoning_tokens":1771,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T16:56:11.371753+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run FF19O simulations of (E4K4)4 from several very different starting structures, such as an extended chain and a compact coil, and compare the resulting radius-of-gyration distributions and SAXS curves; if the narrow distribution (mean about 14.96 Å, spread about 0.27 Å) widens substantially or no longer matches the experimental scattering, the equilibrium assumption fails.","supporting_citations":[],"review_version":2}