{"id":"a1d394dc-2bb1-435d-99ba-26b8af32b96a","arxiv_id":"2505.01860","paper_version":1,"verdict":"UNVERDICTED","confidence":"HIGH","novelty_score":1.0,"correctness_risk":"low","formal_verification":"none","parameter_count":1,"one_line_summary":"A detailed tutorial for setting up, running, and analyzing REST2 enhanced-sampling simulations of intrinsically disordered proteins, demonstrated on an alpha-synuclein fragment.","lead":"Researchers at Dartmouth wrote a step-by-step tutorial for running replica exchange solute tempering (REST2) simulations of disordered proteins. It covers setup, execution, and analysis using GROMACS and PLUMED, including a fully worked example on a fragment of alpha-synuclein.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Tutorial's setup instructions install GROMACS 2024.3/latest PLUMED, while the example was tested with 2022.5/2.9.0; reproducibility of the REST2 protocol is not established.","rationale":"The REST2 theory presented in the paper is standard and the equations are consistent with the literature, so I do not see an internal statistical or formal error. The load-bearing weakness is practical reproducibility: the tutorial's own installation notes install a different GROMACS/PLUMED version combination than the one listed as tested, and the correctness of the demonstration depends on that toolchain's implementation of topology scaling and replica-exchange acceptance. This is a concrete internal inconsistency rather than a hypothetical external bug, which is why I only partially agree with the reader's weakest-assumption framing. A direct side-by-side run with the two version sets would settle whether the mismatch matters. If it does not, the tutorial is usable as written; if it does, the authors need to align the instructions or state that the protocol is version-specific. I therefore recommend a conditional status pending that verification rather than outright acceptance or rejection.","tokens_in":19320,"tokens_out":28047,"duration_ms":283501,"concrete_test":"Run the provided α-syn-C-term REST2 example twice with the same 10-rung ladder: once with GROMACS 2022.5 and PLUMED 2.9.0 as stated in Materials, and once with GROMACS 2024.3 and the PLUMED version installed by Note 1. Compare swap acceptance ratios, mean round-trip times, and base-replica Rg and helical-content distributions with their pyblock errors. If the runs agree within the reported errors, the mismatch is not load-bearing; if they differ, the installation instructions must be corrected or the protocol declared version-specific.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The chapter's central claim is a working recipe for REST2 simulations. This depends on the PLUMED partial_tempering script and GROMACS HREMD machinery implementing Eq. 5's scaling and Eq. 8's acceptance criterion in the same way as in the authors' runs. The manuscript internally contradicts the tested environment: the Materials section states the tutorial was performed with GROMACS 2022.5 and PLUMED 2.9.0, but Note 1 instructs users to clone the current PLUMED2 repository, set GMX_version=2024.3, and patch that build. Following the notes therefore yields a different toolchain from the one that produced Figures 2-4. REST2 is implemented through topology-level atom-type/charge scaling and GROMACS replica-exchange logic; both have changed across these versions. Without a demonstration that the protocol is version-invariant, a user cannot know whether the acceptance ratios, round-trip times, and base-replica ensembles will be reproduced. This is not a criticism of the external tools; it is an internal inconsistency in the tutorial itself.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript is a methods/tutorial chapter that describes how to set up, perform, and analyze all-atom molecular dynamics simulations of intrinsically disordered proteins (IDPs) using replica exchange with solute tempering (REST2) in GROMACS and PLUMED. The authors review the theory of REST and REST2, including the Hamiltonian scaling in Eq. (5) and the Metropolis acceptance criterion in Eq. (8), and then provide a step-by-step workflow: generating starting structures, solvating and equilibrating replicas, creating scaled topology files with PLUMED's partial_tempering script, running the replica-exchange simulation, and analyzing temperature and demultiplexed replicas. The accompanying GitHub repository provides input files, starting structures, and analysis scripts. The example simulation is a 20-residue C-terminal fragment of α-synuclein with 10 replicas spanning an effective solute temperature range of 300–450 K, and Figures 2–4 show temperature diffusion and structural analyses (radius of gyration, helical content, intramolecular contacts).","tokens_in":19533,"tokens_out":10078,"duration_ms":95029,"significance":"If the protocol is reproducible, this manuscript is a valuable community resource for IDP simulation, as it consolidates the REST2 theory, setup commands, and analysis strategies into a single tutorial with freely available input files and scripts. The theoretical section is accurate: Eqs. (5) and (8) correctly match the published REST2 formulation, and the workflow is internally coherent. The authors also provide practical diagnostics (round-trip times, per-replica probability of visiting the base temperature, blocking analysis) that are useful for assessing convergence. The main risk is that the exact commands and software versions as printed in the notes may not reproduce the example, which for a methods chapter is a load-bearing issue for the central claim of a working recipe.","major_comments":[{"comment":"The Materials section states that the tutorial was performed with GROMACS 2022.5 and PLUMED 2.9.0, but Note 1 instructs users to clone the current PLUMED2 repository and to build GROMACS with GMX_version=2024.3. A user following Note 1 therefore obtains a different toolchain from the one that produced Figures 2–4. Since REST2 relies on PLUMED's partial_tempering topology scaling and GROMACS's replica-exchange machinery, both of which have changed across these versions, the protocol's reproducibility is not established as written. Please align the installation instructions with the tested versions, or re-run the example with the newer toolchain and report the resulting acceptance ratios and round-trip statistics.","section":"Materials / Note 1"},{"comment":"The installation commands in Note 1 contain concrete errors that will prevent a following user from building the software: the branch name `v$GMX_version$` has a stray `$`; the line `echo ' export PLUMED_ROOT=$HOME/opt '` is not redirected to `.bashrc` and is missing the `export`; and the `PLUMED_KERNEL` path `/usr/share/lib/libplumedKernel.so` does not match the `$HOME/opt` prefix used earlier in the same note. These instructions should be corrected.","section":"Note 1"},{"comment":"The awk command used to stride replica_index.xvg contains a syntax error (`$1 ==0 $`) and the remaining logic prints the time-0 line and then a line with the time shifted by −80 ps (e.g., the 80 ps entry becomes 0 ps), producing duplicate frame times rather than a clean 80 ps spaced index. Since this command is the basis for constructing demultiplexed trajectories in the analysis workflow, it must be corrected, or the intended code from the repository should be transcribed faithfully.","section":"Note 9"},{"comment":"The manuscript does not state the total production simulation time per replica used to generate Figures 2–4. Without this information, a reader cannot reproduce the example or judge whether the reported convergence diagnostics correspond to a practically achievable run length. Please report the simulated time per replica (and the aggregate time across replicas) used for the example.","section":"Methods / Running REST2 Simulations"}],"minor_comments":[{"comment":"There is a typographical error: \"Hamlitonion\" should be \"Hamiltonian\" in the paragraph preceding Eq. (5).","section":"Theory"},{"comment":"The manuscript contains numerous typos, including \"containg\", \"simulatiosn\", \"uncoverged\", \"choosen\", \"seperately\", \"disucssion\", and \"the the\" in the Software section. These should be corrected in a final pass.","section":"Throughout"},{"comment":"The MDP filename appears as both \"minimz.mdp\" and \"minimiz.mdp\"; please use a single consistent name in the main text and notes.","section":"Notes 5–7"},{"comment":"The name \"Flyvberg\" should be \"Flyvbjerg\" when referring to the blocking analysis; this appears in the analysis section and in the bibliography entry [41].","section":"Simulation Analysis"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a tutorial chapter, and the issues identified are fixable within the manuscript's scope. The GitHub repository itself was not examined; this assessment is based on the manuscript text. The version mismatch and the command errors in the notes are the main barriers to reproducibility and should be addressed before acceptance. The journal's methods/protocol scope is a good fit."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a methods tutorial, not a new methods paper, and that is the right frame. It restates REST2 from Wang et al. 2011, gives the scaling and acceptance equations correctly, and provides a step-by-step example on an alpha-synuclein fragment with input files and analysis code on GitHub. For a newcomer setting up REST2 for an IDP, this is genuinely useful. The theory is sound as far as I checked: Eq. 5 matches REST2, Eq. 8 is the standard exchange acceptance criterion, and the statement that water-water terms vanish from the exchange probability is correct. The practical guidance on starting structures, box size, equal water counts, replica diffusion, and demultiplexing is concrete and sensible. Credit where due: a complete working pipeline is a real contribution.\n\nThe main soft spot is internal and fixable: the Materials section says the tutorial was tested with GROMACS 2022.5 and PLUMED 2.9.0, but Note 1 tells users to clone current PLUMED2 and set GMX_version=2024.3. The acceptance ratios, round-trip times, and base-replica ensembles in Figures 2-4 were generated with the older toolchain; whether REST2 behaves identically in 2024.3 is not shown. For a tutorial whose point is reproducibility, that is a real gap. It is not a failure of the external tools, just an inconsistency the authors should fix by pinning the versions or demonstrating version invariance. Minor typos and a slightly rushed description of blocking analysis are minor.\n\nThe GitHub repository could not be independently verified here, so code-level reproducibility rests on the authors' description. The citation pattern is appropriate; the self-citations track the force field and example system, which is fair.\n\nWho is this for? Graduate students and postdocs who need to run REST2 on an IDP and want a copyable recipe. It delivers that. It is not a research contribution, but it deserves serious peer review because technical details matter. Recommendation: engage, and require the version inconsistency to be resolved before publication.","headline":"A practical, correctly derived REST2 tutorial for IDP simulations, with a real—but fixable—reproducibility gap between the tested software versions and the installation instructions.","tokens_in":700,"tokens_out":1398,"would_cite":false,"duration_ms":26898,"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":"REST2 makes all-atom sampling of disordered proteins practical.","keywords":["intrinsically disordered proteins","molecular dynamics","replica exchange","solute tempering","REST2","enhanced sampling","simulation convergence","alpha-synuclein"],"falsifier":"Take the tutorial's α-synuclein 121–140 system and compare the radius-of-gyration distribution and secondary-structure populations of the 300 K base replica against a long unbiased conventional MD run using the same force field and water model; agreement within the blocking error bars would confirm the REST2 implementation. A cheaper internal check is that exchange acceptance between adjacent rungs stays near or above 20% and that every demultiplexed replica makes round trips from the base rung to the top rung rather than lingering at 300 K.","tokens_in":19124,"feed_emoji":"🧬","tokens_out":11003,"duration_ms":96130,"temperature":0.7,"pith_summary":"This chapter is a worked recipe, not a new theory: it walks through setting up, running, and analyzing all-atom molecular dynamics simulations of intrinsically disordered proteins (IDPs) with replica exchange solute tempering, in the REST2 variant. Its claim is that REST2 makes such simulations practical by heating only the solute through selective Hamiltonian scaling, so far fewer replicas are needed than in ordinary temperature replica exchange. The equations for the scaled Hamiltonian and the exchange acceptance criterion are stated explicitly, and the full pipeline is demonstrated on a 20-residue disordered fragment of α-synuclein. The practical payoff is a reproducible protocol whose 300 K base replica is intended to sample the unbiased Boltzmann ensemble at a fraction of the usual computational cost.","feed_headline":"REST2 makes disordered-protein simulations practical","feed_subtitle":"A step-by-step tutorial shows how to run and verify all-atom REST2 sampling on a disordered protein fragment.","key_machinery":"The central object is the REST2 Hamiltonian, $E_n^{\\text{REST2}} = \\frac{\\beta_n}{\\beta_0} E_{pp} + \\sqrt{\\frac{\\beta_n}{\\beta_0}} E_{pw} + E_{ww}$, where the effective solute temperature is set through $\\beta_n/\\beta_0 = T_0/T_n$; this is what lets a replica thermostated at 300 K behave as though its solute were at 600 K. Its companion is the exchange acceptance criterion, $\\Delta_{n,n+1} = (\\beta_n - \\beta_{n+1})\\left[E_{pp}(X_{n+1}) - E_{pp}(X_n) + \\frac{\\sqrt{\\beta_0}}{\\sqrt{\\beta_n} + \\sqrt{\\beta_{n+1}}}\\left(E_{pw}(X_{n+1}) - E_{pw}(X_n)\\right)\\right]$, in which the water–water energy does not appear. These equations determine the scaled topology for each rung of the solute-temperature ladder and the probability that neighboring replicas swap coordinates. The convergence analysis then turns on demultiplexed replicas—trajectories that follow labeled coordinate sets as they diffuse up and down the ladder—compared through blocking error bars.","core_discovery":"On its own terms, the chapter establishes that the REST2 variant of replica exchange supports a complete, practical all-atom IDP simulation workflow. The solute–solute energy is scaled by $\\beta_n/\\beta_0$, the solute–solvent energy by $\\sqrt{\\beta_n/\\beta_0}$, and water–water interactions are left unscaled, which removes the solvent–solvent energy from the exchange acceptance criterion and lets a logarithmic ladder of effective solute temperatures be spanned with far fewer replicas than temperature replica exchange requires. The paper argues that with equilibrated boxes, scaled topologies built from one unified topology, and exchange attempts every 1.6 ps, the base replica at 300 K samples the unbiased ensemble, and it shows convergence diagnostics—rung-wise averages, demultiplexed-replica round trips, and blocking error estimates—on residues 121–140 of α-synuclein.","pith_inferences":["If the protocol is correct, the same machinery could be applied to a disordered domain within a larger folded protein by limiting the scaled 'solute' region to the disordered segment; the effective-temperature interpretation would then need re-derivation.","The paper's reported high-temperature collapse of IDPs under REST2 suggests a direct three-way test: compare per-rung radius of gyration for REST2, REST3, and tREMD at matched effective temperatures; the discussion predicts the strongest compaction at the top rungs under REST2 alone.","The blocking-error comparison between demultiplexed replicas could be converted into an automated stopping criterion—stop production when all replicas' mean properties agree within one pooled error bar—turning the post hoc convergence judgment into a rule.","The requirement that every replica contain the same number of water molecules could be relaxed by rescaling box dimensions instead, but the paper's equal-box protocol is the safer default because the exchange statistics assume matched solvent degrees of freedom."],"forward_implications":["A single processed topology file can be scaled mechanically into replica-specific topologies for the whole solute-temperature ladder, so preparing a REST2 run does not require rebuilding each replica from scratch.","Because water–water energies are absent from the exchange criterion, adjacent rungs overlap enough that a 300–450 K ladder needs roughly an order of magnitude fewer replicas than tREMD would need for the same range.","The 300 K trajectory from a well-mixed REST2 run is the object to compare with experimental IDP observables, provided the chosen force field and water model are known to describe disordered states.","High exchange acceptance rates alone do not prove convergence; the paper's demultiplexed-replica and blocking analyses are the checks that make the base-replica statistics meaningful.","The same setup generalizes to the REHT and REST3 variants, which modify the scaling to counteract the collapse of IDPs at high solute temperatures."],"supporting_citations":[{"why":"Introduces replica exchange with solute tempering, the Hamiltonian decomposition into solute, solute–solvent, and solvent terms that REST2 builds on.","marker":"[14]"},{"why":"Defines REST2's scaling and exchange acceptance criterion, the exact equations the tutorial implements.","marker":"[15]"},{"why":"Supplies the flexible Hamiltonian replica exchange implementation in the MD engine that lets REST2 run as parallel replicas.","marker":"[16]"},{"why":"Provides the enhanced-sampling plugin and the partial_tempering tool used to generate scaled topologies.","marker":"[19]"},{"why":"Documents the current version of that plugin used by the tutorial workflow.","marker":"[20]"},{"why":"Supplies the a99SB-disp force field and water model that give accurate IDP ensembles in the example.","marker":"[4]"},{"why":"Provides the α-synuclein 121–140 fragment used as the tutorial's demonstration system.","marker":"[25]"},{"why":"Supplies the blocking error analysis used to judge convergence of the simulated structural properties.","marker":"[41]"}],"fun_headline_variants":["REST2: IDP sampling without the replica flood","Cut replicas, not corners: REST2 for IDPs","REST2 streamlines all-atom IDP simulation","Fewer replicas, full ensemble: REST2 for IDPs"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the enhanced-sampling plugin's partial_tempering script and the MD engine's replica-exchange machinery implement the REST2 scaling and acceptance criterion exactly as written; if those external tools carry a bug, or are used with mismatched versions, the 300 K replica will not sample the unbiased Boltzmann ensemble and every tutorial result is void.","fun_headline_variants_meta":{"raw":{"variants":["REST2: IDP sampling without the replica flood","Cut replicas, not corners: REST2 for IDPs","REST2 streamlines all-atom IDP simulation","Fewer replicas, full ensemble: REST2 for IDPs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000624,"raw_usage":{"total_tokens":2846,"prompt_tokens":860,"completion_tokens":1986,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":476,"completion_tokens_details":{"reasoning_tokens":1916}},"tokens_in":476,"tokens_out":1986,"duration_ms":13347,"temperature":1.0,"reasoning_tokens":1916,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T04:08:15.488330+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the tutorial's α-synuclein 121–140 system and compare the radius-of-gyration distribution and secondary-structure populations of the 300 K base replica against a long unbiased conventional MD run using the same force field and water model; agreement within the blocking error bars would confirm the REST2 implementation. A cheaper internal check is that exchange acceptance between adjacent rungs stays near or above 20% and that every demultiplexed replica makes round trips from the base rung to the top rung rather than lingering at 300 K.","supporting_citations":[{"cited_title":"Friesner, and B","cited_arxiv_id":null,"evidence_quote":"Introduces replica exchange with solute tempering, the Hamiltonian decomposition into solute, solute–solvent, and solvent terms that REST2 builds on."},{"cited_title":"Friesner, and B","cited_arxiv_id":null,"evidence_quote":"Defines REST2's scaling and exchange acceptance criterion, the exact equations the tutorial implements."},{"cited_title":"Hamiltonian replica exchange in gromacs: a flexible implementation","cited_arxiv_id":null,"evidence_quote":"Supplies the flexible Hamiltonian replica exchange implementation in the MD engine that lets REST2 run as parallel replicas."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the a99SB-disp force field and water model that give accurate IDP ensembles in the example."}],"review_version":1}